Osteonecrosis in Adult Survivors of Childhood Cancer: A Report From the Childhood Cancer Survivor Study

Abstract

Purpose

Osteonecrosis (ON) is a potentially serious complication of therapy in survivors of childhood cancer. Our goals were to describe the incidence of ON and identify patient and treatment characteristics associated with elevated risk.

Patients and Methods

The rate of self-reported ON was determined for 9,261 patients enrolled onto the Childhood Cancer Survivor Study, a cohort of 5-year survivors of childhood cancer diagnosed from 1970 to 1986, and compared with the rate in a random sample of 2,872 siblings of survivors. Survivors with positive responses were reinterviewed to confirm the diagnosis.

Results

Fifty-two cancer survivors reported ON in 78 joints, yielding 20-year cumulative incidence of 0.43% and a rate ratio (RR) of 6.2 (95% CI, 2.3 to 17.2) compared with siblings, adjusted for age and sex; 44% developed ON in a previous radiation field. The RR was greatest among survivors of stem-cell transplantation for acute lymphoblastic leukemia (ALL), acute myelogenous leukemia (AML), and chronic myelogenous leukemia (RR = 26.9, 66.5, and 93.1, respectively). Nontransplantation patients with ALL (RR = 6.5; 95% CI, 2.2 to 19.4), AML (RR = 11.2; 95% CI, 2.1 to 61.2), and bone sarcoma (RR = 7.3; 95% CI, 2.0 to 26.2) were at higher risk for ON. Older age at diagnosis, shorter elapsed time, older treatment era, exposure to dexamethasone (± prednisone), and gonadal and nongonadal radiation were independently associated with ON.

Conclusion

ON among long-term survivors of childhood cancer is rare. However, compared with siblings, childhood cancer survivors have a significantly increased relative rate of ON, particularly those who were older at diagnosis and who received dexamethasone or radiation therapy. Future studies are needed to better delineate our findings, particularly the increased risk after gonadal radiation.

INTRODUCTION

Currently, approximately 80% of children with cancer will be cured and become long-term survivors. ¹ Osteonecrosis (ON) is an extremely rare disorder in the general population that can be a devastating outcome of therapy in children with cancer. The pathophysiology has not been well elucidated, but the end result is necrosis of one or more bone sites, usually at weight-bearing joints. ² Other terms for ON include avascular necrosis and aseptic ON. Patients experience a spectrum of symptoms that range from mild discomfort to decreased mobility, severe pain at rest, joint swelling, and articular collapse. ³ Some young childhood cancer survivors require surgical joint decompression or replacement.

Previously identified risk factors for ON include glucocorticoid therapy, radiation, and adolescent age at cancer diagnosis. ^4-8 Early case series recognized an association between radiation of the femoral head or jaw and ON. ⁹ However, ON has predominantly been studied and reported in patients exposed to substantial doses of glucocorticoids, often concurrently with other therapies. These studies include mainly children with acute lymphoblastic leukemia (ALL), non-Hodgkin's lymphoma, and history of stem-cell transplantation. There is a paucity of data on the risk of ON in patients with nonhematologic conditions and the independent contribution of other patients and treatment factors.

The published incidence of ON varies widely in children with cancer ^3-5,10-14 and even among studies restricted to patients with ALL. Mattano et al ⁴ reported that 9.3% of 1,409 children with high-risk ALL treated between 1989 and 1995 developed ON within 3 years. In contrast, Arico et al ¹³ found the 5-year cumulative incidence to be 1.6% in ALL patients treated with intensive chemotherapy between 1995 and 1999 based on a chart review for symptoms suggestive of ON. The different findings may be a result of differences in the cumulative dosages of glucocorticoids and/or differences in the types of glucocorticoids (ie, dexamethasone v prednisone). ¹⁰ The toxicity burden of dexamethasone is important to elucidate because of the increased use of dexamethasone in current therapeutic regimens for hematologic malignancies based on promising data from recent clinical trials. ^15,16

Past studies of ON, although provocative, generally have included small patient samples, have focused on one cancer diagnosis, or were based on single-institution experiences. The Childhood Cancer Survivor Study (CCSS) is a 26-site retrospective cohort study of 14,363 childhood cancer survivors, with a wide distribution of cancer diagnoses and treatment exposures. Because of the availability of detailed, high-quality data on chemotherapy and radiation treatment exposure for participants, the CCSS cohort provides a unique sample to further study ON. In this study, we seek to describe the overall incidence of ON among a diverse group of cancer survivors and to identify patient and treatment factors associated with an elevated risk of ON.

PATIENTS AND METHODS

Patients

The CCSS is a multisite, retrospectively ascertained cohort with prospective follow-up designed to study the late effects of childhood cancer therapy. Inclusion criteria include diagnosis of leukemia, CNS malignancy, Hodgkin's disease, non-Hodgkin's lymphoma, malignant kidney tumor, neuroblastoma, soft tissue sarcoma, or bone tumor; diagnosis and initial treatment at one of the 26 collaborating centers between January 1, 1970 and December 31, 1986; age younger than 21 years at diagnosis; and alive 5 years from the date of diagnosis.

Beginning in August 1994, participants completed an extensive baseline questionnaire about their demographic, medical, and psychosocial status. Three subsequent follow-up surveys have been administered thus far to obtain updated information. For participants who returned a signed medical release, data were abstracted from the medical record regarding their initial cancer treatment, treatment for any relapse, and preparatory regimens for bone marrow transplantation. Data regarding exposure to 42 chemotherapeutic agents were abstracted from the medical record; cumulative doses were abstracted for 22 of these agents. Cumulative data for glucocorticoids were not available, and glucocorticoid history for nononcologic conditions was not abstracted. Data were also obtained on the site of the tumor and on fields and doses of radiation therapy. The study design and cohort characteristics have been described previously, ¹⁷ and further details, together with the study documents, are available at http://www.stjude.org/ccss. All CCSS protocol and contact documents were reviewed and approved by the human subjects committee at each participating institution.

Of the 20,691 patients eligible for participation, 17,633 were successfully located, and 14,363 (81.5%) completed the baseline questionnaire. Of the 14,363 initial participants, 9,261 completed the second follow-up questionnaire, which included questions relating to the occurrence of ON. A random sample of participating survivors (n = 6,005) was asked to contact their sibling closest in age for participation in the study; 3,839 siblings completed the baseline questionnaire, and 2,872 siblings completed the second follow-up questionnaire, including the questions regarding ON.

Assessment of ON

Participants were asked if they were ever diagnosed with ON or “avascular necrosis… a condition in which blood supply to the bone or joints becomes interrupted, causing that part of the bone to die,” the date of diagnosis, and what treatment, if any, they received. Survivors who reported a history of ON were administered a brief additional survey by telephone clarifying the exact location(s) of the joints affected, how the diagnosis was made, and the presence of current symptoms and limitations related to the ON. Questions regarding the symptom burden of ON were derived from the Functional Study Index instrument. ¹⁸

Data Analysis

Body mass index was calculated as weight (kg)/height (m)² as described by the Centers for Disease Control and Prevention, with a body mass index ≥ 25 corresponding to overweight status. Because hormone levels were not available, threshold radiation doses were used as surrogates, consistent with previous studies. Radiation ≥ 24 Gy to the hypothalamic- pituitary region was used to categorize patients at high risk for low growth hormone production. ^19-21 Radiation of ≥ 10 Gy ²² to the ovary and ≥ 20 Gy to the testes ²³ was used to define hypogonadism in females (ie, ovarian failure) and males (ie, Leydig cell insufficiency), respectively.

The distribution of the reported sites of ON among patients and their symptoms was described, including the prevalence of pain and limitations of ambulation. Survivors with and without ON were compared in terms of various individual and treatment characteristics using the t test for continuous variables or Fisher's exact test for categoric variables. The cumulative incidence of ON was calculated for survivors and siblings separately and by age at diagnosis among survivors. Patients who did not have ON and died before the second follow-up (1%) were censored. The relative rates of ON in childhood cancer survivors, overall and by cancer diagnosis, were calculated relative to the siblings using Poisson regression adjusting for age and sex.

Among survivors, Poisson regression was used to assess the association of patient and treatment factors with ON rates. We initially performed an unadjusted analysis for each of the patient and treatment factors of interest, followed by a multiple Poisson regression including factors that were marginally significant in the unadjusted analysis (P < .2). We tested a priori hypothesized interactions between age at diagnosis and sex, age at diagnosis and glucocorticoid exposure, and age at diagnosis and gonadal radiation. All statistical analyses were conducted using SAS Version 9.1 (SAS Institute, Cary, NC) and R 2.4.1 (http://www.R-project.org).

RESULTS

Description of ON Among Survivors

Overall, within the cohort, 69 individuals reported a history of ON; 58 successfully completed the additional clarification survey, resulting in 52 confirmed diagnoses of ON. A survivor was classified as having ON only if the additional telephone survey yielded confirmatory information. Survivors with ON were more likely to be older at cancer diagnosis and have a history of allogeneic stem-cell transplantation and exposure to both chemotherapy and radiation (Table 1).

Table 1.

Characteristics of Survivors With and Without Osteonecrosis

Characteristic	Survivors With Osteonecrosis (n = 52)			Survivors Without Osteonecrosis (n = 9,209)			P
		No.	%	No.	%
Age at cancer diagnosis, years					< .001
0-9	13	25	5,820	63
10-15	24	46	2,225	24
16+	15	29	1,164	13
Sex					.064
Female	19	37	4,551	49
Male	33	63	4,658	51
Race					.900
White	45	86	7,883	86
Nonwhite	7	13	1,291	14
Diagnosis					< .001
Acute lymphoblastic leukemia
No transplantation	16	31	2,635	29
Allogeneic transplantation	2	4	62	1
Autologous transplantation	0	0	18	0
Acute myelogenous leukemia
No transplantation	2	4	192	2
Allogeneic transplantation	2	4	31	0
Autologous transplantation	0	0	5	0
Chronic lymphoblastic leukemia
No transplantation	0	0	160	1
Allogeneic transplantation	2	4	21	0
Autologous transplantation	0	0	3	0
CNS tumor	2	4	1,164	13
Hodgkin's lymphoma	9	17	1,173	13
Non-Hodgkin's lymphoma	5	10	692	8
Kidney cancer	0	0	867	9
Neuroblastoma	1	2	624	7
Soft tissue sarcoma	5	10	810	9
Bone sarcoma	6	11	752	8
Stem-cell transplantation					< .001
Allogeneic	6	12	114	1.2
Autologous	0	0	25	0.3
None	46	88	9,070	98.5
Elapsed time since diagnosis,* years					.193
16-20	24	46	3,274	36
21-25	17	33	3,070	33.3
26+	11	21	2,865	31.3
Therapy					< .001
Surgery only	0	0	655	8
Chemotherapy only	0	0	581	7
Radiation only	0	0	24	< 1ff
Chemotherapy + radiation	8	16	1,032	12
Chemotherapy + surgery	5	10	1,598	19
Surgery + radiation	3	6	1,053	13
Chemotherapy + radiation + surgery	35	69	3,445	41

^*Time between diagnosis and second follow-up questionnaire.

Of those affected, 60% had ON in more than one joint (Table 2). The most common site was the hips, followed by the shoulders and knees. ON occurred at the site of previous radiation in 24 survivors (44%). The ON caused difficulty with at least one activity of daily living in 57% of survivors and at least two activities in 44% of survivors. Thirty-three percent of survivors with ON had difficulty walking inside, and 41% currently had pain at rest in the affected bones.

Table 2.

Distribution and Symptoms of Osteonecrosis Observed in 52 Survivors

Characteristic of Osteonecrosis	No. of Survivors With Osteonecrosis	%
Site*
Hip(s)	38	72
Shoulder(s)	13	24
Knee(s)	11	21
Ankle(s)	9	17
Jaw(s)	3	6
Elbow	3	6
Spine	1	2
No. of affected sites
1	21	40
2	14	26
3	10	19
≥ 4	7	13
At least one site in previous radiation field
Yes	23	44
No	19	36
Unknown	10	20
Years between cancer diagnosis and osteonecrosis
0-4	18	35
5-14	16	31
15+	18	35
Osteonecrosis currently causing difficulty with:†
Walking inside	17	33
Climbing up stairs	18	35
Rising from a chair	20	38
Putting on pants	11	21
Reaching into low cupboards	18	35
Opening containers	9	17
Osteonecrosis currently causing difficulty with:
At least one activity from list above	30	58
At least two activities from list above	23	44
How osteonecrosis was diagnosed†
Plain x-ray	49	98
Magnetic resonance imaging	31	69
Bone scan	25	53
History of pain at rest at affected site(s)
Yes	21	41
No	30	59
Missing	1	2

^*Expressed as patients with osteonecrosis reported at that site; for example, 38 patients (72% with a history of osteonecrosis) had osteonecrosis of the hip. Patients may have more than one affected site.

^†Participants may have indicated more than one response.

Rate of ON

The cumulative incidence of ON is 0.43% at 20 years for survivors compared with 0.03% for siblings (Fig 1). The incidence increases with older age at diagnosis (Fig 2), with rates of 0.2% for patients diagnosed at younger than 10 years old (P = .0013), 0.8% for patients diagnosed at 10 to 15 years old (P < .0001), and 1.11% for patients diagnosed at ≥ 16 years old (P = .0001). When ALL patients are examined separately (Appendix Fig A1, online only), the 20-year cumulative incidence is 0.2% in individuals less than 10 years old at diagnosis (P = .02) compared with 2.8% in patients ≥ 16 years old (P = .039).

Fig 1.

Cumulative incidence of osteonecrosis among survivors and a sibling comparison group starting 5 years after diagnosis. The x-axis for a sibling represents elapsed years since the diagnosis date of his/her corresponding survivor.

Fig 2.

Cumulative incidence of osteonecrosis among all survivors stratified by age at diagnosis.

Fig A1.

Cumulative incidence of osteonecrosis among acute lymphoblastic leukemia patients stratified by age at diagnosis.

Overall, survivors were 5.6 times more likely to have reported ON (Table 3) compared with the sibling comparison group, after adjusting for age and sex. The rate ratio (RR) was greatest among survivors of allogeneic stem-cell transplantation for all leukemia subtypes (ALL, acute myelogenous leukemia [AML], and chronic lymphoblastic leukemia: RR = 26.9, 66.5, and 93.1, respectively). Among nontransplantation patients, patients with a history of ALL (RR = 6.5; 95% CI, 2.2 to 19.4) and AML (RR = 11.2; 95% CI, 2.1 to 61.2) had elevated RRs, but the RRs were less than those reported after transplantation. Among nonhematologic malignancies, history of bone sarcoma (RR = 7.3; 95% CI, 2.0 to 26.2) or soft tissue sarcoma (RR = 5.8; 95% CI, 1.6 to 21.8) was significantly associated with ON.

Table 3.

Rates of Osteonecrosis in Survivors Compared With Sibling Comparison Group Controlled for Sex and Age at Evaluation and Stratified by Diagnostic Category and Transplantation Status

Diagnosis and Transplantation Status	Rate Ratio	95% CI	P
All cancers	5.6	2.0 to 15.6	< .001
Diagnosis
Acute lymphoblastic leukemia
No transplantation	6.5	2.2 to 19.4	< .001
Stem-cell transplantation	26.9	4.9 to 147.6
Acute myelogenous leukemia
No transplantation	11.2	2.1 to 61.2	.005
Stem-cell transplantation	66.5	12.2 to 363.5	< .001
Chronic lymphoblastic leukemia
No transplantation	0	0 to 15.6	.99
Stem-cell transplantation	93.1	17.0 to 509.0	< .001
CNS tumor	1.8	0.3 to 9.7	.508
Hodgkin's lymphoma	6.7	2.0 to 22.2	.002
Non-Hodgkin's lymphoma	6.7	1.8 to 25.1	.005
Kidney cancer	0	0 to 300	.99
Neuroblastoma	1.7	0.2 to 15.4	.629
Soft tissue sarcoma	5.8	1.6 to 21.8	.009
Bone sarcoma	7.3	2.0 to 26.2	.002
Stem-cell transplantation status
Allogeneic	59.2	16.7 to 210.5	< .001
Autologous	0	0 to 111.0	.99
None	5.0	1.8 to 14.0	.002

Predictors of ON: Comparisons Among Survivors

Unadjusted Poisson regression identified the following factors as associated with ON (Table 4): older age at diagnosis, glucocorticoid therapy, alkylator therapy, methotrexate, history of any radiation therapy, radiation to the hypothalamic-pituitary region, radiation to the gonads, and older treatment era. Alcohol intake and body mass index were not associated (data not shown).

Table 4.

Determinants of Osteonecrosis Among Survivors Using Unadjusted and Adjusted Poisson Regression Analysis

Factor	Unadjusted Analysis					Adjusted Analysis*
		Rate Ratio	95% CI	P	Rate Ratio	95% CI	P
Sex
Female	0.6	0.3 to 1.0	.068	0.8	0.4 to 1.5	.439
Male	1.0			1.0
Diagnosis
Acute lymphoblastic leukemia
No transplantation	3.5	0.8 to 15.2	.095
Allogeneic/autologous transplantation	15.5	2.2 to 109.9	.006
Acute myelogenous leukemia
No transplantation	6.1	0.9 to 43.2	.071
Allogeneic/autologous transplantation	36.0	5.1 to 255.6	< .001
Chronic lymphoblastic leukemia
No transplantation	0	0 to 11.0	.99
Allogeneic/autologous transplantation	54.3	7.7 to 385.8	< .001
Hodgkin's lymphoma	4.2	0.9 to 19.3	.068
Non-Hodgkin's lymphoma	4.1	0.8 to 21.1	.092
Kidney cancer	0	0 to 2.2	.99
Neuroblastoma	0.9	0.1 to 9.9	.929
Soft tissue sarcoma	3.4	0.7 to 17.4	.146
Bone sarcoma	4.5	0.9 to 22.1	.067
CNS tumor	1.0
Race
Nonwhite	1.0	0.4 to 2.2	.955
White	1.0
Body mass index
≥ 25	0.9	0.6 to 1.6	.852
< 25	1.0
Age at diagnosis, years
≥ 16	5.6	2.7 to 11.8	< .001	5.9	2.7 to 12.8	< .001
10-15	4.7	2.4 to 9.3	< .001	4.5	2.3 to 8.9	< .001
0-9	1.0			1.0
Time since diagnosis, years
≥ 15	0.6	0.3 to 1.1	.105	0.6	0.3 to 1.3	.212
5-14	0.4	0.2 to 0.9	.018	0.5	0.2 to 0.9	.021
0-4	1.0			1.0
Treatment era
1970-1974	2.7	1.2 to 6.5	.022	3.3	1.3 to 8.8	.015
1975-1979	1.3	0.5 to 3.4	.637	1.5	0.5 to 4.3	.464
1980-1986	1.0			1.0
Glucocorticoid history
Dexamethasone ± prednisone	4.0	1.8 to 8.9	< .001	2.7	1.2 to 6.4	.019
Prednisone only	1.7	0.9 to 3.1	.086	1.5	0.8 to 2.9	.189
None	1.0			1.0
Alkylator history
Yes	5.0	2.3 to 10.8	< .001
No	1.0
Methotrexate history
Yes	2.3	1.3 to 4.1	.004
No	1.0
Radiation to pituitary,† Gy
≥ 24	1.1	0.6 to 2.2	.765
< 24	1.0
Radiation to ovary, Gy
≥ 10	2.0	0.6 to 6.4	.246	9.6	2.2 to 42.5	.003
< 10	1.0			1.0
Radiation to testes, Gy
≥ 20	8.3	3.6 to 19.6	< .001	22.2	6.4 to 76.8	< .001
< 20	1.0			1.0
Radiation‡
Yes	4.3	1.7 to 10.8	.002	3.8	1.5 to 9.6	.006
No	1.0			1.0

^*Adjusted Poisson regression model includes all potential contributing factors marginally significant (P < .2) in unadjusted analysis after eliminating diagnosis, history of alkylating history, and methotrexate because of colinearity with corticosteroid history.

^†Used to define patients with low growth hormone production.

^‡In adjusted analysis, this represents nongonadal radiation.

Factors found to be at least marginally significant in univariate analysis were then examined in multiple regression analysis (Table 4). Because glucocorticoid therapy was highly correlated with diagnosis, methotrexate exposure, and alkylator therapy, these latter factors could not be included in the regression model. In the adjusted analysis, the radiation variables were reclassified as nongonadal and gonadal radiation because pituitary radiation was not significantly associated with ON. Stem-cell transplantations were examined in aggregate because a large proportion of the autologous patients had received total-body irradiation (TBI).

Older age at diagnosis, glucocorticoid therapy, gonadal radiation, and shorter elapsed time since diagnosis remained independently associated with ON in multivariate analysis after adjusting for sex and treatment era. Dexamethasone therapy, with or without prednisone, conferred a higher RR of ON than prednisone alone (2.7; 95% CI, 1.2 to 6.4 v 1.5; 95% CI, 0.5 to 4.3). In testing for possible interactions, the association between age at diagnosis and rate of ON did not vary according to glucocorticoid exposure, sex, or gonadal radiation. Similarly, the association between glucocorticoid exposure and ON did not vary with sex.

DISCUSSION

In this large, multisite, retrospective cohort study, childhood cancer survivors were 6.2 times as likely as a sibling comparison group to report the diagnosis of ON. This corresponds to an overall cumulative incidence of 0.43% at 20 years from diagnosis and a 2.8% cumulative incidence for patients 16 years and older. Patients experienced considerable morbidity from their ON, with 57% having at least one limitation among activities of daily living. To our knowledge, this is the largest study to date to examine the risk of ON among long-term survivors of childhood cancer. Because of our large sample size and the distribution of different treatment and patient characteristics, we could examine the role of contributing factors more broadly. We found that children with a history of leukemia, lymphoma, and sarcoma are at the highest risk of developing ON. We confirmed that a history of stem-cell transplantation, radiation exposure, and glucocorticoid therapy and adolescent age at diagnosis are major risk factors for the development of ON. Furthermore, we determined that the type of glucocorticoid formulation is a determinant of risk; patients receiving dexamethasone were approximately 30% more likely to develop ON than patients who received prednisone alone. After adjusting for glucocorticoid history and age, we identified gonadal radiation as a potential independent risk factor in both males and females. Moreover, with the benefit of extended follow-up, we found that the cumulative incidence increases with time for years after treatment. Previous studies have described ON to be an acute effect of therapy and generally not reported after more than the first few years from exposure. ^3,12,24

The cause of the markedly elevated relative rate of ON among recipients of allogeneic stem-cell transplantation has been specifically correlated with long-term glucocorticoid therapy, which is used to treat chronic graft-versus-host disease (GVHD). In a case-control study of 43 children after allogeneic transplantation, Faraci et al ²⁵ found that both TBI and GVHD were independently associated with ON in multivariate analysis. In a retrospective analysis by Socié et al ⁸ of 4,388 patients who had undergone an allogeneic bone marrow transplantation, ON was associated with GVHD but not radiation. Later, Socié et al ²⁶ confirmed these findings in 316 patients with chronic myeloid leukemia and 172 patients AML who had been randomly assigned to either cyclophosphamide (CY) and TBI or busulfan and CY. In multivariate analysis, the relative risk of ON was increased among patients with chronic GVHD and not CY/TBI. These latter data suggest that glucocorticoid therapy is a more important determinant of ON than radiation exposure among transplantation patients.

Although hypogonadism and ON are both possible treatment-related complications after cancer, to the best of our knowledge, hypogonadism has not been previously reported as a risk factor for ON. Our observation that gonadal radiation is an independent risk factor for ON has at least two potential explanations. First, it is possible that the ON is a direct result of radiation-induced damage to a bone/joint within the radiation field (eg, the hip). ^27,28 Alternatively, this could be an indirect effect of radiation-induced gonadal damage and insufficient production of sex steroids. This latter explanation is supported by the increased RR of ON associated with alkylator therapy, another gonadotoxic therapy, in unadjusted analysis. We know that both estrogen and testosterone deficiencies are associated with reduced bone density. ^29,30 We hypothesize that reduced gonadal sex steroids could also affect the risk of ON by a similar or separate mechanism(s). This finding will need to be verified in future studies that include hormonal measurements.

The majority of past studies included only patients with ALL. Overall estimates for the incidence of ON have ranged from approximately 1% ^3,31 to 9%, ⁴ when based on clinical presentation, to approximately 15%, when based on magnetic resonance imaging (MRI) screening. ⁷ Caution should be used when comparing our risk estimates based on patients treated from 1970 to 1986 to more recently treated patients. Because our study relied on self-report, as opposed to the medical record or MRI screening, we may have underestimated the incidence of ON. However, our rate may be a reasonable reflection of the true burden of disease, rather than acute toxicity, because the major morbidities resulting from ON are unrelieved discomfort and decreased ambulatory mobility. These are symptoms that are amenable to self-report. ^18,32 Karimova et al ³³ have found that only larger lesions occupying more than 30% of the femoral head volume were associated with joint collapse or need for arthroplasty, suggesting that not all lesions detected by MRI are clinically significant. Furthermore, one would expect any underestimation to be the same across treatment exposures and diagnoses.

Previous investigators have been concerned about possible additional toxicity of dexamethasone on joints compared with prednisone based on clinical trials with different glucocorticoid schedules. Mattano et al ⁴ noted a higher rate of ON in patients who received higher doses of dexamethasone (23% v 16% in patients who received lower doses) in the delayed intensification phase for high-risk ALL, but the results were not statistically significant. Later, Mattano et al ¹⁰ observed lower rates of ON in patients who received discontinuous dexamethasone (10 mg/m²/d on days 0 through 6 and days 14 through 20) versus higher dose continuous dexamethasone (10 mg/m²/d on days 0 through 20). These previous studies, as well as ours, support that dexamethasone is associated with a greater relative risk of ON compared with prednisone.

Our analysis concludes that older age at cancer diagnosis is associated with a higher RR of ON in an incremental fashion. This age association has been noted by others. ^4,6-8,34 Unlike Mattano et al ⁴ and Arico et al, ¹³ females in our large cohort did not have a higher rate of ON, even in analyses restricted to patients with ALL and in the adolescent age range. This may be because our method of ascertainment was patient self-report, rather than prospective physician assessment. However, other investigators have similarly concluded that sex is not an important contributing factor. ^6,35 In past reports, white race ^4,7 and high body mass index ³⁶ were independent risk factors for ON, none of which were significant in our study.

Several limitations of our study need to be kept in mind. The study design was retrospective, not prospective. Although patients in our cohort generally received the same types of chemotherapeutic agents that are currently used, some of the treatments may not be applicable to the experience of children treated with more modern regimens. The schedule and exact dosage information for the glucocorticoids were not available for our analysis. Specifically, it is not known whether survivors received glucocorticoids for nononcologic diagnoses. In addition, the power for detecting factors that modified the rate of ON was limited as a result of the relatively small number of events, despite the large total number of survivors in the CCSS.

From this large, multisite, cohort study, we conclude that ON occurs in childhood cancer survivors at much higher rates than one would expect in a healthy population. Our results further highlight the considerable discomfort and impaired mobility experienced by individuals with ON. Therefore, medical providers caring for cancer survivors should question their patients about pain and mobility problems in joints and monitor for progression and disability. Focused screening is especially important in stem-cell transplantation recipients and patients who were older at diagnosis. Radiation therapy and glucocorticoids, particularly dexamethasone, confer greater risk. Furthermore, our data suggest that patients who received gonadotoxic therapies may also be at high risk of ON; future studies will be required to confirm this latter observation.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The author(s) indicated no potential conflicts of interest.

AUTHOR CONTRIBUTIONS

Conception and design: Nina S. Kadan-Lottick, Yutaka Yasui, Leslie L. Robison, Charles A. Sklar

Financial support: Leslie L. Robison

Administrative support: Leslie L. Robison

Provision of study materials or patients: Lillian R. Meacham, Marilyn Stovall, Leslie L. Robison

Collection and assembly of data: Nina S. Kadan-Lottick, Leslie L. Robison, Charles A. Sklar

Data analysis and interpretation: Nina S. Kadan-Lottick, Irina Dinu, Sue Kaste, Anita Mahajan, Marilyn Stovall, Yutaka Yasui, Leslie L. Robison, Charles A. Sklar

Manuscript writing: Nina S. Kadan-Lottick, Irina Dinu, Karen Wasilewski-Masker, Sue Kaste, Lillian R. Meacham, Anita Mahajan, Marilyn Stovall, Yutaka Yasui, Leslie L. Robison, Charles A. Sklar

Final approval of manuscript: Nina S. Kadan-Lottick, Irina Dinu, Karen Wasilewski-Masker, Sue Kaste, Lillian R. Meacham, Anita Mahajan, Marilyn Stovall, Yutaka Yasui, Leslie L. Robison, Charles A. Sklar

Appendix

The Childhood Cancer Survivor Study (CCSS) is a collaborative, multi-institutional project, funded as a resource by the National Cancer Institute (NCI), of individuals who survived 5 or more years after diagnosis of childhood cancer.

CCSS is a retrospectively ascertained cohort of 20,346 childhood cancer survivors diagnosed before age 21 between 1970 and 1986 and approximately 4,000 siblings of survivors, who serve as a control group. The cohort was assembled through the efforts of 26 participating clinical research centers in the United States and Canada. The study is currently funded by a U24 resource grant (NCI Grant No. U24 CA55727) awarded to St. Jude Children's Research Hospital. Currently, we are in the process of expanding the cohort to include an additional 14,000 childhood cancer survivors diagnosed before age 21 between 1987 and 1999. For information on how to access and use the CCSS resource, visit www.stjude.org/ccss.

Table A1.

CCSS Institutions	Investigators
St. Jude Children's Research Hospital, Memphis, TN	Leslie L. Robison, PhD,* † Melissa Hudson, MD,† ‡ Greg Armstrong, MD†
Children's Hospitals and Clinics of Minnesota, Minneapolis-St. Paul, MN	Joanna Perkins, MD,‡ Maura O'Leary, MD§
Children's Hospital and Medical Center, Seattle, WA	Debra Friedman, MD, MPH,‡ Thomas Pendergrass, MD§
Children's Hospital, Denver, CO	Brian Greffe, MD,‡ Lorrie Odom, MD§
Children's Hospital, Los Angeles, CA	Kathy Ruccione, RN, MPH‡
Children's Hospital, Oklahoma City, OK	John Mulvihill, MD†
Children's Hospital of Philadelphia, Philadelphia , PA	Jill Ginsberg, MD,‡ Anna Meadows, MD†
Children's Hospital of Pittsburgh, Pittsburgh PA	Jean Tersak, MD,‡ A. Kim Ritchey, MD,§ Julie Blatt, MD§
Children's National Medical Center, Washington, DC	Gregory Reaman, MD,‡ Roger Packer, MD†
Cincinnati Children's Hospital Medical Center, Cincinnati, OH	Stella Davies, MD, PhD†
City of Hope, Los Angeles, CA	Smita Bhatia , MD‡
Columbus Children's Hospital, Columbus OH	Amanda Termuhlen, MD,‡ Frederick Ruymann, MD,§ Stephen Qualman, MD,† Sue Hammond, MD†
Dana-Farber Cancer Institute, Boston, MA	Lisa Diller, MD,‡ Holcombe Grier, MD,§ Frederick Li, MD‖
Emory University, Atlanta, GA	Lillian Meacham, MD,‡ Ann Mertens, PhD†
Fred Hutchinson Cancer Research Center, Seattle, WA	Wendy Leisenring, ScD,† ‡ John Potter, MD, PhD† §
Hospital for Sick Children, Toronto, ON	Mark Greenberg, MBChB,‡ Paul C. Nathan, MD,§
International Epidemiology Institute, Rockville, MD	John Boice, ScD†
Mayo Clinic, Rochester, MN	Vilmarie Rodriguez, MD,‡ W. Anthony Smithson, MD,§ Gerald Gilchrist, MD§
Memorial Sloan-Kettering Cancer Center, New York, NY	Charles Sklar, MD,† ‡ Kevin Oeffinger, MD†
Miller Children's Hospital, Long Beach, CA	Jerry Finklestein, MD§
National Cancer Institute, Bethesda, MD	Barry Anderson, MD,† Peter Inskip, ScD†
Riley Hospital for Children, Indianapolis, IN	Terry A. Vik, MD,‡ Robert Weetman, MD§
Roswell Park Cancer Institute, Buffalo, NY	Daniel M. Green, MD† ‡
St. Louis Children's Hospital, St. Louis, MO	Robert Hayashi, MD,‡ Teresa Vietti, MD§
Stanford University School of Medicine, Stanford, CA	Neyssa Marina, MD,‡ Sarah S. Donaldson, MD,† Michael P. Link, MD§
Texas Children's Hospital, Houston, TX	Zoann Dreyer, MD‡
University of Alabama, Birmingham, AL	Kimberly Whelan, MD, MSPH,‡ Jane Sande, MD,§ Roger Berkow, MD§
University of Alberta, Edmonton, Alberta, Canada	Yutaka Yasui, PhD†
University of California–Los Angeles, Los Angeles, CA	Jacqueline Casillas, MD MSHS,‡ Lonnie Zeltzer, MD† §
University of California–San Francisco, San Francisco, CA	Robert Goldsby, MD,‡ Arthur Ablin MD§
University of Michigan, Ann Arbor, MI	Raymond Hutchinson, MD‡
University of Minnesota, Minneapolis, MN	Joseph Neglia, MD, MPH† ‡
University of Southern California, Los Angeles, CA	Dennis Deapen, Dr. PH†
University of Washington, Seattle, WA	Norman Breslow, PhD†
The University of Texas-Southwestern Medical Center, Dallas, TX	Dan Bowers, MD,‡ Gail Tomlinson, MD,§ George R. Buchanan, MD§
The University of Texas M.D. Anderson Cancer Center, Houston, TX	Louise Strong, MD,† ‡ Marilyn Stovall, MPH, PhD,†

Abbreviation: CCSS, Childhood Cancer Survivor Study.

^*Project principal investigator (U24 CA55727).

^†Member of the CCSS Steering Committee.

^‡Institutional principal investigator.

^§Former institutional principal investigator.

^‖Former member of the CCSS Steering Committee.