Fax: (626) 301-8978
Predictors of avascular necrosis of bone in long-term survivors of hematopoietic cell transplantation
Version of Record online: 17 JUN 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 18, pages 4127–4135, 15 September 2009
How to Cite
Campbell, S., Sun, C.-L., Kurian, S., Francisco, L., Carter, A., Kulkarni, S., Parker, P., Karanes, C., Forman, S. J. and Bhatia, S. (2009), Predictors of avascular necrosis of bone in long-term survivors of hematopoietic cell transplantation. Cancer, 115: 4127–4135. doi: 10.1002/cncr.24474
- Issue online: 4 SEP 2009
- Version of Record online: 17 JUN 2009
- Manuscript Accepted: 31 DEC 2008
- Manuscript Revised: 26 NOV 2008
- Manuscript Received: 11 SEP 2008
- National Institutes of Health. Grant Number: R01CA078938
- Lymphoma/Leukemia Society Scholar Award. Grant Number: 2191-02
- avascular necrosis;
- graft-versus-host disease;
- hematopoietic cell transplantation;
- immunosuppressive agents
Avascular necrosis (AVN) is a debilitating condition reported after chronic steroid use. The purpose of this study was to describe the magnitude of risk in individuals who survived ≥1 years after hematopoietic cell transplantation (HCT), and to investigate the role of immunosuppressive agents such as prednisone, tacrolimus (FK506), mycophenolate mofetil (MMF), and cyclosporine (CSA) in the development of AVN after HCT.
Using a retrospective study design, the authors followed 1346 eligible patients for the development of AVN. Cumulative incidence was calculated taking into consideration competing risk from death and disease recurrence. Cox proportional regression techniques were used to identify associated risk factors.
The median age at HCT was 34 years (range, 7 months-69 years), and median length of follow-up for those surviving was 8.2 years. Seventy-five patients developed AVN of 160 joints. The cumulative incidence of AVN at 10 years was 2.9% after autologous HCT, 5.4% after allogeneic matched related donor HCT, and 15% after unrelated donor HCT (P < .001 compared with autologous HCT recipients). For allogeneic transplant recipients, male sex (relative risk [RR], 2.1; 95% confidence interval [95% CI], 1.1-4.0); presence of chronic graft-versus-host disease (RR, 2.2); and exposure to CSA, FK506, prednisone, and MMF rendered patients at increased risk, especially in patients with a history of exposure to ≥3 drugs (RR, 9.2; 95% CI, 2.42-35.24).
Future studies examining the pathogenetic mechanism underlying AVN should help develop targeted interventions to prevent this chronic debilitating condition. Cancer 2009. © 2009 American Cancer Society.
Avascular necrosis (AVN) of the bone is a painful and debilitating condition that develops when the blood supply to the bone is disrupted, usually in areas with terminal circulation. The condition is believed to be the result of vascular compromise, the death of bone and cell tissue, or disruption of bone repair mechanisms.1-4 AVN has been reported after conventional therapy for childhood acute lymphoblastic leukemia (ALL), particularly after exposure to dexamethasone between the ages of 10 and 20 years.5-7
AVN has also been reported as a complication of hematopoietic cell transplantation (HCT), causing significant morbidity and often requiring surgery. Previous studies have identified graft-versus-host disease (GVHD), older age, primary diagnosis of acute leukemia, total body irradiation (TBI)-based conditioning regimens, and steroid therapy as significant risk factors in patients undergoing allogeneic HCT.8-15 However, these studies have been limited by reliance on small cohorts of allogeneic HCT recipients.10, 14, 15 Although a few small studies have examined the possible role of cyclosporine (CSA) in the development of AVN after HCT,9, 16 to our knowledge the role of the more recently used immunosuppressive agents, such as tacrolimus (FK506) and mycophenolate mofetil (MMF), has not been examined to date.
In the current study, we followed 1346 consecutive patients who had undergone HCT at City of Hope National Medical Center (COH) and survived ≥1 years. Our aim was to describe the magnitude of risk of AVN after autologous or allogeneic HCT, and to examine the role of specific immunosuppressive agents in the development of AVN after allogeneic HCT.
MATERIALS AND METHODS
Subjects and Data Collection
A retrospective cohort study design was used. All consecutive patients who had undergone autologous or allogeneic HCT at COH between 1976 and 1997 for a hematologic malignancy or severe aplastic anemia, had survived at least 1 year after transplantation, and were free of AVN at the time of entry into the cohort were included in this study. A Long-term Follow-up (LTFU) data collection form was completed for all patients meeting eligibility criteria. The form captured information beginning 1 year post-transplantation through the date of last contact. Medical records maintained at COH were the primary source of data for completion of the LTFU form. If the date of last hospital/clinic visit recorded in the medical records was not recent, or if there were any unexpected gaps in the patients' history within the time window of interest, a standard protocol was used to identify and contact physicians who were treating patients outside COH to obtain pertinent information. If the physician was not available or unable to provide recent information, the patient was called directly. This method of follow-up ensured that information regarding clinically symptomatic disease (AVN) was captured in an uninterrupted fashion, from 1 year post-HCT to the date of last contact with a healthcare provider. The Human Subjects Committee at COH approved the protocol. Informed consent was provided according to the Declaration of Helsinki.
Information collected on the LTFU form included demographics, disease status, medication, hospitalization, vaccination history, and post-HCT complications including new malignancies, cardiopulmonary dysfunction, renal compromise, neurologic toxicity, AVN, gastrointestinal complications, and cataracts, as well as details regarding GVHD. Data from the LTFU form were merged with data from an institutional database on HCT containing information on conditioning for HCT and GVHD prophylaxis/treatment.
Our cohort consisted of all 1346 consecutive patients who met the eligibility criteria. This cohort was followed through March 2005. From the LTFU database, cases of AVN diagnosed by a healthcare provider after the first year of HCT were identified. The medical records of all patients with AVN were reviewed and confirmation of the diagnosis was collected using diagnostic radiology (plain x-rays, magnetic resonance imaging, bone scan, and/or computed tomography scan) or surgical report(s). Details regarding management of AVN were also collected.
Cumulative incidence, with 95% confidence intervals (95% CIs), for the development of AVN after HCT was calculated taking into consideration competing risk from death and recurrence of primary disease.17 The time at risk was computed from the date of transplantation to the date of onset of first episode of AVN, date of last contact, date of death, or date of recurrence of primary disease, whichever occurred first. Cumulative incidence (with 95% CIs), for surgical intervention for AVN was calculated taking into consideration competing risk from death.17 The time at risk was computed from the date of onset of AVN to the date of surgery, date of last contact, or date of death, whichever occurred first. The Cox proportional hazards regression method was used to estimate relative risks (RRs) and their 95% CIs.18 Univariate analyses for all pertinent variables were first performed to estimate RR individually. Variables examined in the Cox regression model included year of transplantation (1976-1985 [referent group], 1986-1990, and 1991-1997), age at transplantation (<35 years [referent group] and ≥35 years), race (Caucasian [referent group], Hispanic, and others), type of transplantation (autologous HCT [referent group], allogeneic related donor HCT, and unrelated donor HCT), and primary diagnosis (referent group: acute myeloid leukemia [AML], chronic myeloid leukemia [CML], aplastic anemia [AA], and other malignancies; comparison group: acute lymphoblastic leukemia [ALL], non-Hodgkin lymphoma [NHL], Hodgkin lymphoma [HL], and multiple myeloma [MM]). The primary diagnoses were grouped to create a comparison group that would contain patients potentially at a higher risk for AVN because of exposure to steroids for management of their disease. Other variables examined included agents used for conditioning (TBI, cyclophosphamide, melphalan, busulfan, etoposide, cytarabine, and carmustine), presence of or history of chronic GVHD (yes or no), and exposure to immunosuppressive agents used for GVHD prophylaxis and/or treatment. The major drugs used for GVHD prophylaxis/treatment for the study cohort included methotrexate, systemic prednisone, CSA, FK506, and MMF. To examine the role of CSA, systemic prednisone, FK506, and MMF, we combined these agents into the following composite variables: no CSA; exposure to CSA alone (without exposure to prednisone, FK506, or MMF); exposure to CSA and prednisone; exposure to CSA, prednisone, and MMF; exposure to CSA, prednisone, and FK506; and finally, exposure to CSA, prednisone, FK506, and MMF. The majority of patients receiving no CSA received methotrexate alone for GVHD prophylaxis.
Stepwise regression was employed to select important variables from those that approached statistical significance in the univariate analysis. A P value of <.10 was used as the selection criterion. Multivariate Cox regression was then performed for the selected variables to obtain adjusted RRs and 95% CIs. Analyses were conducted for the entire cohort, and stratified by the type of HCT (autologous transplant recipients only and allogeneic transplant recipients only). All tests of statistical significance were 2-sided; and a P value <.05 was considered statistically significant. Data were analyzed with SAS statistical software (version 9.1; SAS Institute Inc, Cary, NC).
The clinical characteristics of the patient population are summarized in Table 1. Of the 1346 patients in this cohort, 793 (59%) were males. The median age at transplantation was 34 years (range, 7 months-69.1 years). As of March 2005, 883 patients (66%) were alive at last contact, and the median length of follow-up for these patients was 8.2 years (range, 1-25.3 years). The major indications for transplantation were AML (n = 335), NHL (n = 299), CML (n = 230), HL (n = 169), ALL (n = 178), MM (n = 61), AA (n = 56), and other miscellaneous diagnoses (n = 18). Six hundred seventy-one (50%) patients received grafts from human leukocyte antigen (HLA)-matched or partially matched family member donors; 81 (6%) patients received grafts from unrelated donors matched for HLA phenotype, and 594 (44%) patients underwent autologous transplantation. The main agents used for conditioning included cyclophosphamide (73%), TBI (75%), etoposide (69%), carmustine (11%), busulfan (9%), melphalan (3%), and cytarabine (5%). Among allogeneic transplant recipients, chronic GVHD was diagnosed in 419 patients (56%). Multiple drugs were used for GVHD prophylaxis and/or treatment, including CSA (82%), methotrexate (66%), prednisone (74%), FK506 (8%), and MMF (4%).
|Characteristic||Total, N=1346, No. (%)||Patients With AVN, n=75, No. (%)|
|Year of transplant|
|1976-1985||177 (13)||6 (8)|
|1986-1990||256 (19)||17 (23)|
|1991-1997||913 (68)||52 (69)|
|Age at HCT, y|
|<35||726 (54)||46 (61)|
|≥35||620 (46)||29 (39)|
|Male||793 (59)||57 (76)|
|Female||553 (41)||18 (24)|
|Caucasian||857 (64)||45 (60)|
|African American||33 (3)||3 (4)|
|Hispanic white||342 (25)||23 (31)|
|Asian||84 (6)||4 (5)|
|Other||30 (2)||0 (0)|
|Acute myeloid leukemia||335 (25)||23 (31)|
|Non-Hodgkin lymphoma||299 (22)||11 (15)|
|Chronic myeloid leukemia||230 (17)||12 (16)|
|Acute lymphoblastic leukemia||178 (13)||10 (13)|
|Hodgkin lymphoma||169 (13)||11 (15)|
|Multiple myeloma||61 (5)||5 (6)|
|Aplastic anemia||56 (4)||2 (3)|
|Other||18 (1)||1 (1)|
|Chemotherapeutic agents used for conditioning|
|Cyclophosphamide||988 (73)||53 (71)|
|TBI||1013 (75)||57 (76)|
|Etoposide||928 (69)||50 (67)|
|Carmustine||147 (11)||8 (11)|
|Busulfan||122 (9)||9 (12)|
|Cytarabine (Ara-C)||72 (5)||1 (1)|
|Type of HCT|
|Autologous||594 (44)||20 (27)|
|Allogeneic||752 (56)||55 (73)|
|Related||671 (50)||44 (59)|
|Unrelated||81 (6)||11 (14)|
|Status at last contact|
|Alive||883 (66)||57 (76)|
|Dead||463 (34)||18 (24)|
|Drugs used for GVHD prophylaxis/treatment (allogeneic HCT only)|
|Cyclosporin A||614 (82)||52 (95)|
|Tacrolimus (FK506)||57 (8)||16 (29)|
|Methotrexate||493 (66)||36 (65)|
|Prednisone||558 (74)||49 (89)|
|Mycophenolate mofetil||31 (4)||9 (16)|
|Chronic GVHD||419 (56)||45 (82)|
Seventy-five patients developed AVN ≥1 years post-HCT. Of these, 44 patients had received grafts from HLA-matched or partially matched family member donors and 11 from unrelated donors, and 20 had received autologous transplants. In total, 160 joints were affected, including 106 hips, 26 knees, 19 shoulders, 2 ankles, 4 wrists, and 3 elbows. The median number of affected joints per patient was 2 (range, 1-8 joints). The median latency from date of transplantation to AVN development was 3.2 years (range, 1 year-16.6 years).
As shown in the top panel of Figure 1, the cumulative incidence of AVN was 3.7% (95% CI, 2.7%-4.7%) at 5 years from HCT and 5.0% (95% CI, 3.7%-6.2%) at 10 years. The bottom panel of Figure 1 shows the cumulative incidence of AVN by type of transplantation. The cumulative incidence was lowest for autologous HCT recipients (2.5% at 5 years and 2.9% at 10 years), intermediate for matched related allogeneic HCT recipients (3.6% at 5 years and 5.4% at 10 years; P = .05 compared with autologous HCT recipients), and highest for unrelated HCT recipients (13.2% at 5 years and 14.7% at 10 years; P < .001 compared with autologous HCT recipients). Among allogeneic HCT recipients, the cumulative incidence was 2.8% for females and 5.8% for males at 5 years post-HCT (P = .01). The corresponding figures for autologous HCT recipients were 1.2% for females and 3.5% for males (P = .06).
Of the 75 patients with AVN, 40 underwent surgery for management of AVN. The cumulative incidence of surgery was 30.9% (95% CI, 20.3%-41.5%) at 1 year from AVN diagnosis (Fig. 2). Arthroplasty was needed to replace 64 affected joints (40% of all affected joints); an additional 9 joints (5.6%) required either drilling or placement of pins.
Risk Factors for AVN: Entire Cohort
Table 2 presents the results of the univariate analyses for the association between AVN and demographic and clinical factors, and drugs used for prophylaxis and/or treatment of GVHD. For the entire cohort, univariate analyses identified HCT performed after 1985, male sex, and allogeneic transplantation as risk factors for development of AVN. Exposure to specific chemotherapeutic agents or TBI used for conditioning was not found to be associated with an increased risk of AVN (data not shown). The final multivariate analysis (Table 3) revealed matched related (RR, 2.4; 95% CI, 1.30-4.36 [P < .001]) or unrelated (RR, 5.8; 95% CI, 2.58-13.11 [P < .001]) donor HCT, HCT performed after 1985, and male sex (RR, 2.3; 95% CI, 1.33-3.85 [P = .002]) to be independently associated with an increased risk of AVN after HCT.
|Factor||Total||Autologous HCT||Allogeneic HCT|
|No. With/ Without AVN||RR (95% CI)||No. With/ Without AVN||RR (95% CI)||No. With/ Without AVN||RR (95% CI)|
|Year of transplant|
|1986-1990||17/239||2.65 (0.99-7.09)||5/97||1.00||12/142||3.00 (1.07-8.40)|
|1991-1997||52/861||2.73 (1.08-6.92)||15/477||0.70 (0.25-1.93)||37/384||4.00 (1.55-10.35)|
|Age at transplant, y|
|≥35||29/591||0.86 (0.54-1.37)||9/363||0.52 (0.22-1.27)||20/228||1.32 (0.76-2.29)|
|Male||57/736||2.35 (1.39-4.00)||15/323||2.32 (0.84-6.37)||42/413||2.33 (1.25-4.35)|
|Hispanic white||23/319||1.19 (0.72-1.96)||7/88||2.74 (1.08-6.97)||16/231||0.77 (0.42-1.39)|
|African American/Asian/other||7/140||0.90 (0.41-1.99)||1/57||0.62 (0.08-4.76)||6/83||0.87 (0.36-2.07)|
|Type of transplant|
|Related donor||44/627||1.59 (0.93-2.71)||44/627||1.00|
|Unrelated donor||11/70||4.30 (2.06-8.98)||11/70||2.74 (1.40-5.33)|
|ALL/NHL/HL/MM||37/670||1.02 (0.65-1.60)||18/465||2.14 (0.50-9.23)||19/205||1.37 (0.78-2.38)|
|Drugs used for GVHD prophylaxis/treatment (allogeneic HCT only)|
|CSA alone||3/68||2.33 (0.47-11.67)|
|CSA + prednisone||31/443||4.17 (1.24-14.05)|
|CSA + prednisone + MMF||2/8||12.38 (2.02-75.94)|
|CSA + prednisone + FK506||9/25||18.42 (4.85-69.98)|
|CSA + prednisone + FK506 + MMF||7/14||19.77 (4.99-78.36)|
|P for trend||<.001|
|Chronic GVHD (allogeneic HCT only)|
|Factor||Overall, RR (95% CI)||Allogeneic, RR (95% CI)|
|Type of transplant|
|Related donor||2.39 (1.30-4.36)||1.00|
|Unrelated donor||5.82 (2.58-13.11)||1.67 (0.82-3.43)|
|Year of transplant|
|Male||2.26 (1.33-3.85)||2.10 (1.12-3.95)|
|ALL/NHL/HL/MM||1.55 (0.93-2.59)||1.58 (0.90-2.77)|
|Drugs used for GVHD prophylaxis/treatment|
|CSA alone||—||2.26 (0.45-11.32)|
|CSA + prednisone||—||2.62 (0.74-9.28)|
|CSA + prednisone + MMF||—||5.17 (0.78-34.39)|
|CSA + prednisone + FK506||—||9.83 (2.34-41.23)|
|CSA + prednisone + FK506 + MMF||—||7.98 (1.76-36.18)|
|P for trend||—||<.001|
|Presence of chronic GVHD|
Risk Factors for AVN by Type of Transplantation
For autologous transplant recipients, no independent risk factors were identified. For allogeneic HCT survivors, in the univariate analysis, HCT performed after 1985, male sex, unrelated donor HCT, presence of chronic GVHD, and combinations of ≥3 drugs used for GVHD prophylaxis/treatment were associated with increased risk of AVN (Table 2). In the multivariate analysis for allogeneic HCT recipients, male sex, combinations of ≥3 drugs used for GVHD prophylaxis/treatment, and the presence of chronic GVHD were associated with increased risk of AVN (Table 3). Compared with women, men had a higher risk of AVN (RR, 2.1; 95% CI, 1.12-3.95 [P = .01]). The presence of chronic GVHD (RR, 2.2; 95% CI, 1.02-4.82 [P = .05]) and history of exposure to CSA, FK506, prednisone (PSE), and MMF independently rendered patients at an increased risk of AVN, especially in patients exposed to ≥3 drugs (RR, 9.2; 95% CI, 2.42-35.25 [P = .001]). Because the number of patients exposed to MMF was relatively small, we analyzed the data after excluding these patients. Exposure to CSA, FK506, and PSE together rendered patients at an increased risk of AVN (RR, 8.9; 95% CI, 2.13-37.46 [P = .003]) when compared with patients not exposed to CSA.
Of the 1346 patients who underwent HCT and survived at least 1 year free of the outcome of interest, we found the cumulative incidence of AVN to be 3.7% at 5 years and 5% at 10 years after transplantation. Unrelated donor HCT recipients were at increased risk of developing AVN compared with autologous HCT recipients. Among allogeneic HCT recipients, the risk of AVN was increased in male patients, those diagnosed with chronic GVHD, or those exposed to at least 3 of the following drugs: prednisone, CSA, FK506, and MMF.
AVN results from impairment of circulation to the affected bone. The femoral head is affected most frequently. Systemic steroids have long been attributed to predisposing a patient to AVN,19-21 either by the development of fat emboli in the microvasculature of the affected bone or by fatty infiltration of the bone marrow. Increased intraosseous pressure, intravascular coagulation, venous stasis, and hyperviscosity syndrome have also been ascribed a role in the development of AVN.1 The role of intravascular coagulation is supported by the finding that the factor V Leiden gene, a risk factor for coagulation, is over-represented in patients with AVN when compared with controls.22
To the best of our knowledge, there is limited information regarding the role of calcineurin inhibitors (CSA and FK506) and MMF in the development of AVN after HCT. Although the association between calcineurin inhibitors and AVN has been studied by others,9, 16, 23-25 some studies have implicated both CSA and FK506 as risk factors,25 whereas others have implicated only CSA.24, 26 However, research has shown that both CSA and FK506 are thrombogenic in patients with hematologic malignancies, an observation that supports the association between exposure to these agents and the development of AVN.27-29 Furthermore, prior studies have also demonstrated an additive effect with exposure to CSA and FK506 for the development of both hyperlipidemia30 and thrombotic microangiopathy.31 This additive effect may be a basis for the observations of the current study that risk of AVN was increased in patients exposed to prednisone, CSA, FK506, and MMF, with the risk increasing with the number of agents used.
Chronic GVHD has been reported as a risk factor for the development of AVN. However, because of the strong correlation between the use of systemic prednisone and the presence of chronic GVHD, it is sometimes difficult to attribute the risk of AVN to chronic GVHD alone. There is emerging evidence that indicates that chronic GVHD may play an independent pathogenetic role in the development of AVN, due to the increased risk of microangiopathy.32, 33 As described above, immunosuppressive agents such as CSA and FK506 have also been reported to cause vascular damage,33, 34 thus setting the stage for an increased risk of AVN associated with both factors through the same mechanism. Our results support the hypothesis that chronic GVHD and immunosuppressive agents are independent risk factors for the development of AVN.
The results of previous studies have been inconsistent when describing the sex-related differences in risk of AVN. Some studies have reported an increased risk among males,11 whereas others have demonstrated an increased risk in females,10 and some have failed to demonstrate a sex preference.8, 15 The results of the current study demonstrated an increased risk of AVN in males, primarily in allogeneic HCT recipients, which could potentially be explained by differential metabolism of calcineurin inhibitors by sex, an area that needs further exploration.
The current study describes the low magnitude of risk of AVN after autologous HCT. The cumulative incidence of AVN in autologous HCT recipients approached 2.5% at 5 years. The large majority of events were diagnosed by 5 years from HCT. No clinical or demographic risk factors were identified.
The current study has some limitations. AVN patients were diagnosed after they developed clinical signs and symptoms, and not as a result of routine radiologic screening. Our results therefore may not be generalizable to those diagnosed through routine radiologic screening. Furthermore, it would have been ideal to have the dose and duration of the immunosuppressive agents, to understand more completely the individual contributions of the immunosuppressive agents in the development of AVN. However, it is logistically impossible to calculate the dose or duration of the individual immunosuppressive agents used for the long-term management of chronic GVHD in this population, and therefore this study could not explore a detailed dose-response association. The observed increased risk among those transplanted after 1986 is most likely a reflection of the changing transplantation practices, in particular the increasing number of unrelated donor HCTs, and the use of a wider variety of immunosuppressive agents for the management of chronic GVHD.
The strengths of this study include a comprehensive and complete assessment of a large population of HCT recipients, including autologous HCT recipients, and the ability to examine the role of immunosuppressive agents such as calcineurin inhibitors, MMF, and prednisone, independent of chronic GVHD in the development of AVN after allogeneic HCT. AVN is a debilitating condition, frequently requiring surgical interventions to ensure relief from pain, as evidenced by the finding that 30% of the patients underwent surgery by 1 year from diagnosis, and approximately 50% required surgery by 5 years. Future studies should examine the pathogenetic mechanism underlying AVN, thereby setting the stage for targeted interventions to prevent this chronic debilitating condition.
Conflict of Interest Disclosures
Supported in part by National Institutes of Health grants R01 CA078938 (to S.B.) and P01 CA30206 (to S.J.F.), and the Lymphoma/Leukemia Society Scholar Award for Clinical Research 2191-02 (to S.B.).
- 18Regression models and life tables. J R Stat Soc. 1972; B: 187-220..
- 28Causes of acute thrombotic microangiopathy in patients receiving kidney transplantation. Exp Clin Transplant. 2004; 2: 268-272., , , et al.
- 29Drug-related thrombosis in hematologic malignancies. Rev Clin Exp Hematol. 2004; 8: E4., , .