Fractures among long-term survivors of childhood cancer

A report from the Childhood Cancer Survivor Study




Although reductions in bone mineral density are well documented among children during treatment for cancer and among childhood cancer survivors, little is known about the long-term risk of fracture. The objective of this study was to ascertain the prevalence of and risk factors for fractures among individuals participating in the Childhood Cancer Survivor Study (CCSS).


Analyses included 7414 ≥5-year survivors of childhood cancer diagnosed between 1970 and 1986 who completed the 2007 CCSS follow-up questionnaire and a comparison group of 2374 siblings. Generalized linear models stratified by sex were used to compare the prevalence of reported fractures between survivors and siblings.


The median ages at follow-up among survivors and siblings were 36.2 years (range, 21.2-58.8 years) and 38.1 years (range, 18.4-62.6 years), respectively, with a median 22.7 years of follow-up after cancer diagnosis for survivors. Approximately 35% of survivors and 39% of siblings reported ≥1 fracture during their lifetime. The prevalence of fractures was lower among survivors than among siblings, both in males (prevalence ratio, 0.87; 95% confidence interval, 0.81-0.94; P < .001) and females (prevalence ratio, 0.94; 95% confidence interval, 0.86-1.04; P = .22). In multivariable analyses, increasing age at follow-up, white race, methotrexate treatment, and balance difficulties were associated with increased prevalence of fractures among female survivors (P = .015). Among males, only smoking history and white race were associated with an increased prevalence of fracture (P < .001).


Findings from this study indicated that the prevalence of fractures among adult survivors did not increase compared with that of siblings. Additional studies of bone health among aging female cancer survivors may be warranted. Cancer 2012. © 2012 American Cancer Society.


Survivors of childhood cancer are at risk of developing bone-related late effects as a result of disturbances in bone metabolism during childhood or adolescence. Attainment of normal peak bone mass may be compromised by the effects of the cancer experience, such as nutritional deficiencies and reduced exercise capacity,1, 2 or because normal bone mineral accretion and skeletal development are affected by corticosteroids and other chemotherapeutic agents (eg, methotrexate).2-4 Bone mineral density (BMD) can be adversely affected by gonadal failure after exposure to radiation or gonadotoxic chemotherapy or as a consequence of hypothalamic pituitary dysfunction after irradiation to the central nervous system (CNS).5, 6 Moreover, direct radiation to bone causes cytotoxic effects on the epiphyseal chondrocytes, increased hypervascularity, and reduced bone strength.7-9

Deficits in BMD among survivors of childhood cancer have been well documented in numerous studies.9-12 Among children receiving treatment for acute lymphoblastic leukemia (ALL), it has been observed that bone mass decreases significantly during treatment13-15 followed by a return to lower than average value (although still within the normal range) in the years after the completion of therapy.16 In the general population, reduced BMD is a major public health concern, as decrements in BMD can significantly increase the risk of fracture, which in turn is associated with elevated rates of disability and mortality and high socioeconomic costs.17-19 Accordingly, failure to accrue sufficient bone mass during childhood and adolescence may increase the risk for early onset osteoporosis among childhood cancer survivors and can place them at risk for fracture later in life. Despite this, the occurrence of fracture among long-term survivors remains largely uncharacterized. Previous studies that have measured fracture risk among survivors have examined only fractures occurring during treatment or within the first 5 years of completing therapy.13, 15 Thus, it is not clear whether alterations to bone metabolism during therapy have an impact on post-therapy risk of fractures. Moreover, most studies of fracture risk among survivors have been restricted either to individuals who were treated previously for ALL or malignancies of the CNS15, 20 or to small sample sizes that limited the consideration of additional factors, such as demographic and lifestyle factors, on fracture risk.

The objectives of this study were to describe the history of reported fractures among a large and diverse cohort of cancer survivors and to identify treatment-related and host-related factors that predispose survivors to an increased risk of fracture. A major advantage of this study is the availability of detailed information on treatment, health-related behaviors, physical activity levels, body mass index, and balance and movement disorders, which may predispose survivors to increased risk for fracture as they age.


The Childhood Cancer Survivor Study (CCSS) is a multisite, retrospective cohort designed to study the late effects of childhood cancer therapy. Potential study participants were identified from 26 participating institutions across the United States and Canada based on the following criteria: diagnosis of leukemia, CNS malignancy, Hodgkin lymphoma, non-Hodgkin lymphoma, malignant kidney tumor, neuroblastoma, soft tissue sarcoma, or bone tumor; diagnosis date between January 1, 1970 and December 31, 1986; age <21 years at diagnosis; and alive 5 years from the date of diagnosis. Information relating to each study participant's original cancer diagnosis and treatment was abstracted from medical records held at participating centers. The CCSS study protocol and cohort characteristics have been described previously21, 22 (available at:; accessed November 15, 2011). All CCSS protocol and contact documents were reviewed and approved by the human subjects committee at each participating institution.

Of the 20,691 survivors of childhood cancer who were eligible for participation, 17,633 were successfully contacted, and 14,358 completed the baseline questionnaire. Of the 14,358 initial participants, 8013 completed the 2007 follow-up questionnaire (Fig. 1). A random sample of participating cancer survivors (n = 6100) was asked to contact their sibling closest in age for participation in the study. Of these, 4828 stated willingness to participate and were sent a questionnaire. Of these, 4023 siblings completed the baseline questionnaire, and 2374 siblings completed the 2007 follow-up questionnaire.

Figure 1.

This flow diagram illustrates study participation of cancer survivors from the Childhood Cancer Survivor Study.

The primary outcome measure for these analyses was the occurrence of a fracture among study participants. Self-reported data from the 2007 follow-up questionnaire were used to characterize fracture history. Participants were asked whether they had “ever broken a bone.” If the participant's response to this question was “yes,” then he/she was asked to provide further details regarding their previous fracture(s).

Information about medical conditions or functional limitations potentially associated with fracture risk and about physical activity, height, and weight also were obtained from the 2007 questionnaire. Participants were asked to report whether they had ever been diagnosed with problems affecting their balance or equilibrium and to grade the extent of their problem from mild (not affecting walking or daily routine) to disabling. Respondents who reported a moderate-to-disabling problem were considered to experience difficulties with balance and equilibrium. Vision loss was defined as a diagnosis of legal blindness in 1 or both eyes. Smoking history was categorized as ever versus never. Body mass index was calculated as weight (kg) divided by height squared (m2) and was classified as underweight (<18.5 kg/m2), normal (20-24.9 kg/m2), overweight (25-29.9 kg/m2), and obese (≥30 kg/m2). Physical activity was classified as meeting (yes/no) the Centers for Disease Control and Prevention guidelines for physical activity (30 minutes of moderate-intensity physical activity 5 or more days a week or 20 minutes of vigorous activity 3 or more days a week). Functional ability was determined based on the participant's responses to 5 questions, which were adapted from the National Health Interview Survey, that asked participants whether their health over the past 2 years was limited for more than 3 months in 1) the type or amount of moderate activities they can do, like moving a table, carrying groceries, or bowling; 2) uphill walking or climbing up a few flights of stairs; 3) bending, lifting, or stooping; 4) walking 1 block; or 5) eating, dressing, or bathing.23

In addition to health status, selected chemotherapeutic agent exposure (methotrexate and alkylating agents), as well as glucocorticoids, were considered in analyses based on prior knowledge of the influence of these agents on bone metabolism.2-5 Cumulative doses of glucocorticoids were unavailable. CNS and pelvic radiation exposures and surgical procedures, including amputation of the lower limb (transtibial, transfemoral, or hemipelvectomy) and bilateral orchiectomy and bilateral oophorectomy, were obtained from medical records. Exposures to selected chemotherapeutic agents and pelvic radiotherapy as well as the occurrence of an amputation were classified as dichotomous variables for analyses. Radiation exposure to the hypothalamus and pituitary was categorized as no exposure, between 1 and 2000 centigrays (cGy), or >2000cGy. Bilateral orchiectomy and bilateral oophorectomy were not considered in further analyses because of the small number of patients who had undergone these procedures (n < 20). Finally, data on current use of agents known to promote bone health, including hormone-replacement therapies, bisphosphonates, vitamin D, and calcium supplements, were also considered in analyses.

The demographic and treatment characteristics of survivors participating in the current study were compared against those survivors who did not complete the CCSS 2007 follow-up questionnaire, that is, nonparticipants, using chi-square statistics. Descriptive statistics were calculated for demographic and health characteristics and compared between survivors and siblings. Generalized linear models, stratified by sex, were used to compare the history of fractures between survivors and siblings. Generalized estimating equations with robust variance estimates were used to account for intrafamily correlation. The results are presented as prevalence ratios with 95% confidence intervals (CIs). The influence of selected factors on the history of fractures only among survivors also was evaluated using generalized linear models (log-link with binomial distribution) stratified by sex. Factors that were considered in the analysis included smoking status, body mass index, physical activity, vision loss, difficulties with balance and equilibrium, functional health status, and treatment type. Only those variables that demonstrated P values < .2 in bivariate models were considered in multivariable models. All models were adjusted for ethnicity, attained age, and age at diagnosis. All analyses were performed using the statistical package SAS version 9.2 (SAS Institute, Inc., Cary, NC).


The median ages at follow-up among cancer survivors and siblings were 36.2 years (range, 21.2-58.8 years) and 38.1 years (range, 18.4-62.6 years), respectively. The median age at diagnosis was 6.9 years (range, 0-21 years) for survivors, and their median length of follow-up was 22.7 years (range, 15.6-34.2 years). Among both cancer survivors and their siblings, the majority of participants were of white, non-Hispanic descent (>89%) (Table 1). Compared with siblings, survivors were less likely to have ever smoked (P < .001). Compared with nonparticipants, survivors were more likely to be female (50.6% vs 42.1%; P < .001) and of white, non-Hispanic descent (90.8% vs 79.9%; P < .001). In addition, a higher proportion of survivors had received glucocorticoids (46.8% vs 34.9%) or methotrexate (43.6% vs 31.4%) than nonparticipants. It is noteworthy that data regarding chemotherapeutic exposures were unavailable for approximately 30% of nonparticipants.

Table 1. Demographic Characteristics of Survivors and Siblings Who Completed the Childhood Cancer Survivor Study 2007 Follow-Up Questionnaire
 Survivors (n = 7414)Siblings (n = 2374) 
  • Abbreviations: BMI, body mass index; CNS, central nervous system; HL, Hodgkin lymphoma; NA, not applicable; NHL, non-Hodgkin lymphoma; STS, soft tissue sarcoma.

  • a

    Chi-square statistic.

Age at follow-up, y     
 White non-Hispanic673490.8211589.1.042
 Black non-Hispanic1812.4542.3 
 Other non-Hispanic1802.4612.6 
 Not specified300.4833.5 
Ever smoker     
 Not determined2543.4431.8 
 CNS malignancy90512.2   
 Bone tumor6078.2   
Alkylating score  NA  
 Dose unknown4476   
Methotrexate  NA  
Glucocorticoid  NA  
 Prednisone only311442.0   
Cranial irradiation  NA  
Pelvic irradiation  NA  

Over a third of survivors (34.8%) and siblings (38.9%) reported the occurrence of 1 or more fractures during their lifetime. The most frequently reported site of a fracture was the upper limb both for survivors (54.9%) and for siblings (55.6%), as indicated in Table 2, followed by fractures of the lower limb and skull. The distribution of the total number of fractures reported by participants did not vary between survivors and siblings (P > .05) (Table 3). After adjusting for attained age, ethnicity, smoking status, body mass index, and history of medications known to promote bone health, male survivors of childhood cancer were less likely to report a fracture than their siblings (prevalence ratio, 0.87; 95% CI, 0.81-0.94; P < .001). Although the reported prevalence of fractures also was lower among female survivors compared with their sibling counterparts (prevalence ratio, 0.94; 95% CI, 0.86-1.04; P = .22), this association did not meet significance at an alpha (α) level of P = .05.

Table 2. The Absolute Number of Fractures Among Survivors and Siblings by Site
  • a

    Chi-square statistic.

  • b

    Includes fractures of the humerus, radius, ulna, scapula, clavicle, carpals, metacarpals, and phalanges of the hand.

  • c

    Includes fractures of the pelvis, femur, tibia, fibula, patella, tarsals, metatarsals. and phalanges of the foot.

  • dInsufficient information was provided by the study participant to determine the site of the fracture.

Upper bodyb250454.989755.6 
Other upper body190649.381049.8 
Lower bodyc153233.648132 
Other lower body137130.144729.7 
Total4564 1565  
Table 3. Distribution of the Total Number of Fractures Reported by Study Participants
No. of FracturesNo.%No.%Pa
  • a

    Chi-square statistic.


Generalized linear models stratified by sex were used to examine the influence of selected characteristics on the prevalence of fractures among survivors of childhood cancer. In multivariable analyses, male survivors of nonwhite ethnic descent were less likely to report a fracture than white participants (prevalence ratio, 0.78; 95% CI, 0.66-0.92; P = .004). Only prior smoking history (prevalence ratio, 1.24; 95% CI, 1.14-1.34; P < .001) was associated with an increased prevalence of fracture (Table 4). Among female survivors, an association between increasing age at follow-up and an increased prevalence of fractures was observed: survivors between ages 40 and 49 years were 1.22 times (95% CI, 1.01-1.48; P = .044) and survivors aged >50 years were 1.48 times more likely (95% CI, 1.10-1.99; P = .009) to report a fracture than survivors between ages 18 and 29 years. Female survivors who reported difficulties with balance or equilibrium (prevalence ratio, 1.25; 95% CI, 1.05-1.48; P = .012) or who had received methotrexate treatment (prevalence ratio, 1.15; 95% CI, 1.03-1.27; P = .001) also reported an increased prevalence of fracture in multivariable analyses.

Table 4. Multivariable Analysis of the Risk of Fractures Among Survivors of Childhood Cancer Stratified by Sexa
VariablePR95% CIPPR95% CIP
  • Abbreviations: CI, confidence interval; PR, prevalence ratio.

  • a

    Only those variables that demonstrated P values <.2 in bivariate models were included in multivariable models.

  • b

    Variable with a statistically significant P value.

  • c

    Individuals who reported 1 or more of the following were classified as experiencing limitations to activity: inability to walk 1 block or participate in moderate physical activity; difficulty climbing a few flights of stairs; difficulty bending, lifting, or stooping; requiring help to dress, eat, and bathe.

  • d

    Cancer survivors who participated in 30 minutes of moderate-intensity physical activity on 5 or more days a week or 20 minutes of vigorous activity 3 or more days a week were classified as meeting the Centers for Disease Control and Prevention guidelines for regular physical activity.

Demographic variables      
 Age at diagnosis, y      
  0-41.0  1.0  
 Age at follow-up, y      
  18-291.0  1.0  
  White1.0  1.0  
 Ever smoker      
  No1.0  1.0  
Treatment-related variables      
 Alkylating agent      
  No1.0  1.0  
 Radiation to the pelvis      
 Amputation of lower limb      
Functional limitations and physical activity variables      
 Balance and equilibrium problems      
 Limitation to activityc      
 Meets guidelines for physical activityd      

Table 5 presents a comparison of the prevalence of fractures among individual cancer diagnostic groupings compared with siblings. Among male survivors, a history of any diagnosis except non-Hodgkin lymphoma and bone tumors was associated with a decreased risk of fracture compared with siblings. The observed prevalence of fracture was reduced significantly only among female survivors of kidney tumors (prevalence ratio, 0.76; 95% CI, 0.62-0.93; P = .009). The only diagnostic group that had an observed higher prevalence of fracture compared with the sibling control group was female survivors of bone tumors (prevalence ratio, 1.15; 95% CI, 0.97-1.36), although this finding was not statistically significant at P = .05.

Table 5. The Risk of Fracture Among Survivors by Childhood Cancer Diagnosis Compared With Siblings
 Males Females
 ≥1 Fracture   ≥1 Fracture   
DiagnosisNo.%PRa95% CIPNo.%PRa95% CIP
  • Abbreviations: CI, confidence interval; CNS, central nervous system,; HL, Hodgkin lymphoma; NHL, non-Hodgkin lymphoma; PR, prevalence ratio; STS, soft tissue sarcoma.

  • a

    The PR was adjusted for attained age and ethnicity.

Siblings109746.61.0  127732.31.0  
 CNS tumor46930.70.660.56-0.76<.00143627.30.840.71-1.00.054
 Bone tumor29343.00.900.78-1.05.1831437.61.150.97-1.36.10


The objectives of this study were to characterize history of fractures in a large cohort of adult survivors of childhood cancer and to identify patient and treatment characteristics associated with fractures. To date, only a limited number of studies have examined the risk of fracture among childhood cancer survivors, and findings have been contradictory.6, 10, 20 In the current study, we observed that the prevalence of reported fractures was comparable between female survivors of childhood cancer and their siblings; whereas, among males, the prevalence of fracture was lower among survivors relative to the sibling control group. Furthermore, we did not observe any meaningful differences in the total number of reported fractures or in the distribution of fractures by site between survivors and their siblings.

Although numerous studies have reported an increased risk of BMD deficits among children who are receiving treatment for cancer or shortly after they finish therapy, the implications of anticancer therapies on bone mass and fracture risk are not clear among long-term survivors. In fact, some investigators have observed that BMD values for many survivors will return to the normal range in the years after the completion of therapy.16 Accordingly, for survivors participating in the current study, BMD may recover sufficiently with time such that their risk of fracture will not increase above that of their siblings. However, it is also important to consider that BMD is not the only factor mediating the risk of fracture, because bone strength also is determined in part by the inherent structural and material properties of bone, including geometry, trabecular thickness and connectivity, cortical porosity, mineral-to-matrix ratio, and collagen composition.24, 25 Although it has been demonstrated that these abnormalities in bone quality and strength increase the propensity to fracture,26-28 little is known about the structural and material properties of bone among adult survivors of childhood cancer and how these properties are affected by anticancer treatments. It is possible that, among survivors with deficits in BMD, reduced bone mass may not be sufficient to increase fracture risk substantially but that corresponding impairment in bone quality and strength also may be necessary. Thus, further studies characterizing additional measures of bone strength and quality (ie, bone geometry) as well as BMD may be important for understanding the etiology of fractures among adult survivors of childhood cancer.

In addition to bone quality and strength, factors that increase the propensity to fall, such as impairments in vision, neuromotor coordination, postural control, and muscle function, also may increase fracture risk. In the current study, we observed an increased risk of fractures among female survivors who reported difficulties with balance and equilibrium consistent with findings from a previous study in breast cancer survivors29 as well as studies in nononcologic populations.30, 31 Among survivors of childhood cancer, difficulties with balance and equilibrium may be the result of therapy with platinum-based agents, which can damage the organs of the inner ear, impairing vestibular function. Damage to the nerve fibers of the hands, legs, and feet also can occur after vincristine or platinum therapy.32, 33 Our findings suggest that the presence of chronic health conditions after therapy for childhood cancer, such as deficits in balance, may increase the risk of fracture among survivors. However, not all chronic health conditions that increase skeletal fragility or the propensity to fall were associated with increased fracture risk in our study.

In the current study, an association between increasing age at follow-up with a greater prevalence of fracture was observed among female survivors. In the general population, rates of fracture begin to rise substantially among women in their 50s, coinciding with declines in estrogen that occur after menopause.24, 25 Previous findings from the CCSS indicate that female survivors of childhood cancer are more likely to enter menopause prematurely compared with their siblings.34 Accordingly, our observation of an increasing prevalence of fracture among aging female survivors may be caused in part by an earlier decline in estrogen production. However, we did not observe an association between exposure to alkylating agents or pelvic irradiation with an increased prevalence of fracture. The absence of an association may be explained in part by hormone-replacement therapies among female survivors who are diagnosed with hypogonadism. It is also possible that failure to reach peak genetic potential during adolescence and early adulthood, because of the effects of disease and anticancer treatments on normal bone accrual, may have promoted the premature onset of age-related fracture among older females in our survivor cohort. Although the underlying reason for our observation is unclear, our finding suggests that further studies of bone health among aging female cancer survivors may be warranted.

A limitation of this study was the use of self-report questionnaires to collect information on the occurrence of fractures and other health-related information among survivors of childhood cancer and their siblings. The reliability of this approach depended on a study participant's ability to report the occurrence of prior fractures; consequently, biases in recall may have impacted our ability to estimate the prevalence of fractures in the study cohort. We are reassured that the validity of self-reports for fractures is high based on previous studies in both men and women.35 This study also was limited by the absence of data on BMD, which prevented us from evaluating potential associations between fracture risk and BMD in the study population. Finally, the greater proportion of females and individuals of white, non-Hispanic descent among survivors who completed the CCSS 2007 follow-up questionnaire may limit our ability to generalize findings to males and to survivors of nonwhite descent. Although we also observed that a greater proportion of participants who received glucocorticoids and methotrexate also completed the CCSS 2007 follow-up questionnaire, these observations are difficult to interpret given the high number of nonparticipants for whom treatment information was unavailable.

Overall, the findings from this study indicate that the prevalence of fracture among long-term adult survivors of childhood cancer is similar to that of their siblings despite chemotherapy and radiation exposure known to disrupt bone metabolism during therapy. Nevertheless, caution is required when interpreting these results because the majority of study participants have yet to reach an age at which the underlying population risk of fracture increases substantially. The long-term trajectory of BMD deficits after anticancer therapies in childhood is poorly defined, and little is known about the potential effects of chemotherapy and radiotherapy on the bone health of ageing survivors. Moving forward, it will be important to characterize long-term skeletal morbidities in postmenopausal and aging childhood cancer populations.


This project was funded by grant U24CA055727 (L. L. Robison, Principal Investigator) from the National Cancer Institute. Additional funding was provided by a St. Jude Children's Research Hospital Cancer Center Support Grant (5P30CA021765-33). Dr. Kadan-Lottick is supported in part by an American Cancer Society Scholar Grant (119700-RSGHP-10-107-01-CPHPS).


The authors made no disclosures.