|Home | About | Journals | Submit | Contact Us | Français|
Osteosarcoma survivors have received significant chemotherapy and have undergone substantial surgeries. Their very long term outcomes (20 year) are reported here.
We assessed the long-term outcomes of 733 five-year survivors of childhood osteosarcoma diagnosed from 1970-1986 to provide a comprehensive evaluation of medical and psychosocial outcomes for survivors enrolled on the Childhood Cancer Survivor Study (CCSS). Outcomes evaluated included overall survival, second malignant neoplasms (SMNs), recurrent osteosarcoma, chronic health conditions, health status(general and mental health and functional limitations), and psycho-social factors. Outcomes of osteosarcoma survivors were compared to general-population statistics, other CCSS survivors, and CCSS siblings.
Survivors had a mean follow-up of 21.6 years. The overall survival of children diagnosed with osteosarcoma who survived 5-years at 20 years from original diagnosis was 88.6% (95% CI: 86.6-90.5%). The cumulative incidence of SMNs at 25 years was 5.4% with a Standardized Incidence Ratio of 4.79 (95% CI: 3.54-6.33, p<0.01). Overall, 86.9% of osteosarcoma survivors experienced at least one chronic medical condition and over 50% experienced two or more conditions. Compared to survivors of other cancers, osteosarcoma survivors did not differ in their reported general health status (OR=0.9, 95% CI: 0.7-1.2), but were more likely to report an adverse health status in at least one domain (OR=1.9, 95% CI: 1.6-2.2) with activity limitations (29.1%) being the most common.
Childhood osteosarcoma survivors in this cohort did relatively well considering their extensive treatment; but are at risk to experience chronic medical conditions and adverse health status. Survivors warrant life-long follow-up.
There are approximately 400 individuals under the age of 20 years diagnosed with osteosarcoma each year in the United States.1 The treatment for patients with osteosarcoma includes the use of multi-agent chemotherapy and complete surgical resection. The type of surgical intervention is dependent on the site, age of the patient, extent of disease, and expertise/preference of the orthopedic surgeon. The 5-year overall survival rates are approaching 70%.2, 3 However, outcomes have been associated with disease and treatment-related sequelae4, which can have a negative impact on the patients’ quality of life and need for continued access to health care services.5
Given that childhood cancer is relatively rare, and young adult survivors do not always return to their treatment center for long-term follow-up, a single institution will not have sufficient numbers of patients to draw firm conclusions about long-term outcomes. The CCSS,6 a multi-institutional study of individuals surviving five or more years following treatment for childhood cancer, provides a unique opportunity to assess the outcomes of long-term survivors of pediatric osteosarcoma. This report provides a comprehensive evaluation of survival and patient related medical and psychosocial outcomes and health status.
The study design and cohort characteristics of CCSS have been previously described in detail.6 The 26 CCSS collaborating institutions initially identified 20,631 eligible five-year cancer survivors diagnosed under the age of 21 years between 1970-1986. 14,357 (69.6%) were successfully located and completed a baseline survey in 1995-96. Additional follow-up surveys were conducted in 2000 and 2003. Copies of surveys are available from www.stjude.org/ccss. Informed consent was obtained from all participants.
A total of 733 (68.1%) of 1,077 eligible osteosarcoma survivors completed the baseline questionnaire. Non-participants included survivors who refused (n=194), were lost-to-follow-up (n=148), or were non-English speakers (n=2). Overall in the CCSS, participants, refusers, and lost-to-follow-up subjects were found similar with regard to the distributions of sex, age at diagnosis, age at the entry to CCSS, diagnosis type, and type of cancer-treatment.6, 7
To compare survivors with cancer-free individuals, a random sample of siblings of the CCSS survivors were selected and asked to complete the CCSS questionnaires. There were 3,899 (81.4%of eligible) siblings who completed the baseline questionnaire, 2,540 siblings and 552 survivors who completed the 2000 follow-up survey and 2,875 siblings and 488 survivors who completed the 2003 follow-up survey. All of the outcomes utilized all available data from the three surveys except for chronic health conditions and adverse health status, which was available only from the baseline questionnaire.
The US National Death Index (NDI) was used to ascertain the vital status of all 1,068 eligible osteosarcoma survivors as of December 31, 2002 with causes of death being identified through death certificates (causes of nine Canadians’ deaths could not be ascertained).8
Treatment information including tumor location was collected from the treating institution. Data regarding exposure to 42 chemotherapeutic agents (yes/no) were abstracted from the medical record; with cumulative doses abstracted for 22 specific agents.6 Cancer related surgical procedures and fields/doses of radiation therapy were also collected. As described previously,9 for alkylating agent, the total dose received per body surface area (mg/m2) was summed for each agent received and the overall distribution of the non-zero doses among all participants divided into tertiles. Among patients exposed to an alkylating agent, the dose score was calculated by adding the tertile score (1, 2, or 3) for each of the alkylating agents given to a particular patient. A score of 0 was assigned to non-exposed patients. The total anthracycline dose and platinum agent dose were scored and grouped into tertiles by total dose levels.10, 11
SMNs and recurrences (local or distant) were ascertained initially through self-report via the questionnaires. Cohort members were asked to report occurrences of any cancer (either relapse or new cancer) since the diagnosis of original cancer. The pathology reports of positive responses were obtained and SMNs validated by the CCSS pathologist.9 Nonmelanoma skin cancers, non-malignant meningiomas, and other nonmalignant CNS tumors were excluded from the SMN analysis. SMNs were coded by histology using the International Classification Diseases for Oncology (ICD-O).12 Diagnostic groupings of SMNs were constructed following the International Classification of Childhood Cancer.13
Chronic medical conditions were assessed for participants of age 18 years and older using the Common Terminology Criteria for Adverse Events.14 As described earlier15 Grades 1 to 4 of chronic medical conditions represent, respectively, mild, moderate, severe, and life-threatening or disabling.
Hearing and cardiovascular outcomes were individually evaluated for incidence and associations with risk factors.
Six domains of health status (general health, mental health, functional status, activity limitations, pain as a result of the cancer or its treatment, and anxiety as a result of the cancer or its treatment) were assessed among survivors who were alive when they completed the baseline questionnaire using the same method as an earlier report.16 Brief Symptom Inventory (BSI-18)17, 18 was used to assess the mental health. Functional status and activity status were classified based on answers to questions adapted from the National Health Interview Survey and the Behavioral Risk Factor Surveillance System Survey Questionnaire.19, 20 To compare osteosarcoma survivors to the other CCSS survivors, Ewing’s sarcoma survivors were excluded because their treatment similarity may have lessened any noted differences. No comparison for osteosarcoma survivors vs. siblings was made for pain and anxiety since the questions on cancer-related pain and anxiety were not asked of siblings.
Psychosocial outcomes were compared among living osteosarcoma survivors and CCSS siblings of age 25 years and older (time at which one generally completes schooling and transition to independence from parents) at the time of the most recent survey. This included marital status, education, employment, and health insurance.
Descriptive statistics for demographic variables and treatment exposures (radiotherapy, chemotherapy, surgery including amputation) were tabulated for the 733 osteosarcoma survivors and the 3,899 siblings who participated in the CCSS study.
Standardized mortality ratios (SMRs) for all cause and cause-specific deaths of 1,068 eligible osteosarcoma survivors, stratified by sex, were calculated using age-sex-calendar-year-specific US mortality rates.21 The person time at risk for death was taken as the time from the cohort entry (5 years from cancer diagnosis) to the earlier of death or the follow-up end date of December 31, 2002 (NDI cutoff date). The Kaplan-Meier method22 was used to estimate survival curve following study entry, and compared with a hypothetical survival curve that assumed the age-sex-calendar-year matched US population mortality rates for the osteosarcoma survivors. To identify factors associated with mortality risk in the 733 osteosarcoma survivors, a multivariable Poisson regression model was used with a pre-specified set of potential risk factors.
The cumulative incidence of SMNs and that of recurrence of original cancer were calculated separately, taking death as a competing risk event. Vital status of the survivors was described in relation to early (within 5 years since osteosarcoma diagnosis) and late (after 5 years post diagnosis) recurrence. SMNs were enumerated by type initially, followed by Poisson regression analysis of Standardized Incidence Ratios (SIRs) for developing SMNs using sex, age at diagnosis, time since diagnosis, and radiation exposures as explanatory variables. The person time at risk for SMN was taken as the time from the cohort entry to the earliest of SMN development, date of death, and the most recent questionnaire completion date. Age-sex-calendar-year-specific cancer incidence rates from the National Cancer Institute’s Surveillance, Epidemiology, and End-Results (SEER) registry23 were used as reference rates in the SIR calculation.
Numbers and percentages of osteosarcoma survivors, CCSS siblings, and other CCSS survivors for chronic medical conditions, health status, and psychosocial outcomes were reported. Multivariable log-binomial regression for relative risks and logistic regression for odds ratios, adjusting for age, sex and amputation were used for comparisons of these outcomes in the osteosarcoma survivors with the CCSS siblings and other CCSS survivors.
Prevalence of tinnitus and/or vertigo, and that of deafness and/or use of hearing aid were tabulated and assessed by multivariable logistic regression. Incidence of cardiovascular outcomes was evaluated using multivariable Poisson regression. Each of these analyses of specific chronic conditions was performed with respect to a set of pre-specified risk factors.
For the osteosarcoma survivors, psychosocial outcomes were tabulated and compared with the CCSS siblings, using polytomous logistic regression, adjusting for age and sex, while taking into account the potential within-family correlation between survivors and siblings using nonparametric bootstrap of families.24
All statistical analyses were conducted with SAS version 9.1.3, using two-sided inference including standard asymptotic 95% confidence intervals (CIs) associated with each regression analysis.
Table 1 lists the characteristics of the osteosarcoma survivor study population and CCSS sibling cohort. The 733 osteosarcoma survivors had a mean age at diagnosis of 13.7 years (2 to 20 years), mean age at last contact of 35.3 years (13 to 51 years), and mean length of follow-up of 21.6 years (5.1 to 33.8 years). At last contact, the sibling cohort was younger than the cases with an average age of 31.3 years (3 to 58 years).
Among 5+ year survivors of osteosarcoma, subsequent survival at 10, 15, and 20 years since diagnosis was 93.5% (95% CI: 92.1-95.0%), 90.4% (95% CI: 88.7-92.2%) and 88.6% (95% CI: 86.6-90.5%), respectively. Figure 1 compares the observed overall survival of the osteosarcoma survivor cohort with overall survival for the age- and sex-adjusted US population. There were gender differences in overall survival (p=0.04). with an advantage for females at 10 years (94.9% vs. 92.4%), 15 years (92.3% vs. 88.9%) and 20 years (90.0% vs. 87.3%).
Of the 1,068 eligible survivors, 138 (12.9%) died following entry into the cohort (Table 2). Seventy-five of those deaths were due to recurrent or persistent osteosarcoma. The overall SMR was 5.85 (95% CI: 4.91-6.91) with females having a higher SMR at 8.40 compared to males at 4.92: the higher SMR in females was due to the lower mortality rates of females, relative to males, in the US population. Significant SMR (Table 2) elevations are seen with SMN and cardiac etiologies. Risk factors for mortality, determined in regression models (data not shown) included exposure to cisplatinum, radiation therapy, and the development of a second malignancy or recurrent disease.
Only 24 first recurrences (local and distant) occurred after 5 years from diagnosis. The cumulative incidence of recurrences at 5, 10, 15, 20 and 25 years were 24.5%, 27.2%, 27.6%, 28.0%, and 28.1%, respectively (Figure 2). In contrast to the SMNs, whose rates continued to increase across time, the majority of recurrences occurred in ten years following diagnosis (Figure 2).
There were a total of 49 SMNs (Table 3)(in 37 survivors) among the 733 osteosarcoma survivors who participated; 61.2% occurred among female survivors. The most common SMNs was female breast cancer. The majority (85.7%) of SMNs occurred after 10 years from diagnosis (Figure 2) and the SIRs were significantly elevated(Table 3) regardless of gender. The cumulative incidence of SMNs at 25 years from OS diagnosis was 5.4% (males 4.1%, females 7.1%) (Figure 2). The SIRs were elevated for female breast cancer, thyroid cancer and GI cancer (Table 3). Multivariable Poisson regression analysis did not identify any particular demographic or treatment related risk factors for developing an SMN. Eight of the 37 patients who were diagnosed with an SMN after entry into the cohort subsequently died
The prevalence of tinnitus/vertigo and deafness/hearing aid requirement at the baseline survey in this group of osteosarcoma survivors was 14.7% and 7.0%, respectively. The prevalence of these complications increased with increasing doses of platinum for both tinnitus and deafness: no platinum (7.8%, 3.7%), platinum dose 1-362mg/m2 (14.6%, 9.8%), platinum dose of 363-489mg/m2 (23.3%, 8.2%), platinum dose of >490mg/m2 (29.6%, 15.7%). After adjusting for gender, age at diagnosis and platinum dose females were more likely to have deafness or need hearing aids (p=0.03), or to report tinnitus or vertigo (p=0.07) than were males. Survivors diagnosed with osteosarcoma at early age (0-9 years) were less likely to report tinnitus or vertigo compared to those who were older (15+) at diagnosis (OR 0.42, 95%CI: 0.19-0.93, p=0.03).
Eleven of the 733 (1.5%) osteosarcoma survivors reported having coronary heart disease more than 5 years from diagnosis. The prevalence of coronary heart disease was similar across all time frames from the time of diagnosis. After adjusting for gender, current age, time since diagnosis and anthracycline cumulative exposure, no significant predictors were identified. Nineteen survivors reported having congestive heart failure more than 5 years from diagnosis. Survivors’ gender, age at diagnosis, time since diagnosis, and anthracycline were not predictive of this complication, although more events occurred in patients who were exposed to higher doses of anthracyclines
Eighty-seven percent of osteosarcoma survivors reported at least one chronic medical condition. Greater than 50% experienced two or more chronic medical conditions and 76.1% experienced Grade-3 or -4 conditions. When compared to siblings, osteosarcoma survivors were more likely to have any grade, multiple grades (≥ 2) and high grades (3 or 4) of chronic medical conditions. Relative to other childhood cancer survivors, osteosarcoma survivors were more likely to have higher grade conditions (RR=1.4, 95% CI: 1.2-1.6, p<0.001), even, after adjusting for amputation status, gender and age. No difference in the prevalence of chronic medical conditions was seen between the osteosarcoma survivors and survivors of other childhood cancers.
More than half of the osteosarcoma survivors 18 years and older report having had some adverse heath outcome compared to one-third of other survivors and 17.7% of siblings. Approximately 10.3% of osteosarcoma survivors reported having adverse general health, 15.8% reported adverse mental health, and 15.6% impaired functional status. Additionally among the osteosarcoma survivors, 29.1% reported a physical limitation and 22.1% reported having pain from their osteosarcoma and its treatment, while 12.2% reported still having anxiety/fears as a result of their diagnosis and treatment. When compared to siblings, osteosarcoma survivors were more likely to have impaired health status across all measures. But when osteosarcoma survivors are compared to other survivors of childhood cancer, osteosarcoma survivors were only more likely to experience functional limitations, activity limitations and pain.
Overall, 25.4% of survivors reported having never been married compared to 20.3% of siblings. 7.2% of survivors did not graduate from high school while 2.7% of siblings did not. Unemployment rates were low, but more survivors (3.1%) were unemployed versus siblings (0.2%). In terms of health insurance, both groups had a significant number without insurance.
Due to the relatively low incidence (approximately 400 per year) of childhood osteosarcoma in the US1, conducting a large comprehensive study of osteosarcoma survivors at any one institution is very difficult. However, the CCSS with its large multi-institutional cohort enables comprehensive assessments of various sequelae of a large number of 5-year+ osteosarcoma survivors. This is an important subset of patients to examine in detail as these patients receive very invasive surgeries and multimodal chemotherapy with known significant toxicities.
Mortality estimates in this cohort of 5-year survivors of osteosarcoma were elevated when compared to the US population with a 20-year survival probability of 88.6%which is similar to the reported overall CCSS mortality25. Females had a slight survival advantage compared to males (90.0% vs. 87.3%), however, females had a higher SMR compared to males, because of the lower mortality rates of females (compared to males) in the US population coupled with the increased risk of an SMN in females. The development of an SMN or recurrent disease were predictive of death, as was the use of radiation and exposure to cisplatinum which may be a marker for extent or aggressiveness of disease. In data not shown, it is noted that radiation therapy was predominately given to older cohorts when survival was likely poorer and systemic therapy was limited. When cisplatinum exposure was further examined, higher cumulative exposure was associated with deaths attributed to recurrences, SMNs, and cardiac toxicity. However, cumulative cisplatinum exposure is correlated with other exposures such as higher cumulative dose of anthracycline. Thus, the exact contribution of cisplatinum to the observed increase in risk of death is not clear, and it is possibly a surrogate for more aggressive disease or disease recurrence necessitating more chemotherapy exposure.
When comparing our survival rates of 5-year survivors with those reported in other studies, caution must be used as the majority of the survivorship studies report survival from diagnosis as compared to survival of those alive 5 years from diagnosis. In one large study of 1,702 patients with newly diagnosed osteosarcoma between 1980 and 1998 from the German Cooperative Osteosarcoma Study Group, the overall survival at five years from diagnosis was 65.3%. Survival rates continued to decline with 10 and 15 years survival rates of 59.8% and 57.3%. These rates would approximate this study’s survival rates of 5 year survivors at 10 and 15 years from diagnosis.2
The cumulative incidence of recurrences in our study was relatively stable after 5 years (24.5% at 5 years and 28.1% at 25 years) with only 24 subjects experiencing a recurrence more than 5 years after diagnosis. This is comparable to a large study of 576 patients with relapsed osteosarcoma where only 5.7% of the relapses occurred after 5 years.26
In contrast to the recurrences, the majority of SMNs occurred ten or more years from diagnosis. A prior report from CCSS showed that bone tumor survivors had a cumulative incidence of 3.3% at 20 years 9. The current report extends these data and focuses on osteosarcoma patients. The SIR for all SMNs was 4.8 with a cumulative incidence of 1.0% at 10 years and 5.4% at 25 years, which was consistent with a 10-year cumulative incidence reported by Goldsby et al27. Several other studies have reported on the incidence of SMNs with rates ranging from 2 to 5% at 10 years28-31.
The literature reports that the incidence and magnitude of hearing loss increases with increasing cumulative cisplatinum doses with a threshold between 240-400 mg/m2 with younger age associated with greater risk.32, 33 In the current study, we confirmed increased hearing loss and increased prevalence of vertigo with increasing cumulative cisplatinum dose. However, younger age at diagnosis was not predictive of this late sequelae, which may be due to the small numbers of children who were diagnosed at a very young age among the osteosarcoma group (12 survivors under the age of 4 years at diagnosis) in this cohort.
Although anthracyclines are known to have cardiac toxicity, they are an integral part of osteosarcoma therapy and the majority of patients with osteosarcoma now receive close to what is considered the maximum cumulative dosing (450 mg/m2). Recent studies34 have demonstrated a cumulative incidence of 10% at 20 years for acute congestive heart failure for those who received over 300 mg/m2, but even those with lower cumulative anthracyclines doses may be at risk.35 In our study, 19 who reported having congestive heart failure with 13 of the 19 having received greater than 360 mg/m2 of anthracyclines. Reports of congestive heart failure are important to note in this cohort, as nearly 86% received anthracyclines with 58.2% receiving greater than 210 mg/m2 of anthracyclines. Congestive heart failure tended to be more common in patients farther from treatment. In addition to congestive heart failure 11 reported having cardiovascular disease. One limitation of this data is that this relies on self reporting which is an underestimation and also it is unknown how well surveillance imaging and follow-up was performed in this group of survivors.
As previously identified, bone sarcoma survivors were more likely than the siblings to have chronic and adverse health conditions.15, 16 Since survivors of osteosarcoma in the current study had a large number of Grade 3 or 4 chronic health conditions due to the high number of amputations (72%) performed in this treatment era (1970-1986), we adjusted for amputations when comparing against siblings and other survivors. This is reflected in decreased relative risks when compared to Oeffinger et al. paper.15 Interestingly osteosarcoma survivors had more Grade 3-4 conditions (even after adjusting for amputation) compared to other survivors, likely reflecting the effects of extensive surgery and chemotherapy received by this group of survivors.
When general health status was examined, osteosarcoma survivors were more likely than the sibling cohort to report adverse health status. When osteosarcoma survivors were compared to survivors of other tumors, general health, mental health and anxiety were not different, but functional status, activity limitations and pain were more likely to be a problem. Given the use of amputation and surgeries in this osteosarcoma cohort compared to other survivors, this would be expected. However, with 72% of the CCSS osteosarcoma survivors having an amputation, one would expect a greater number reporting impaired functional status (15%) or activity limitation (29%). This is remarkable and is likely due to adaptation to their limitations. The increased incidence of chronic conditions and limitations and pain seen in osteosarcoma survivors is especially concerning, given the relative lack of health insurance of this osteosarcoma cohort (25.9%). Survivors of osteosarcoma also seemed to compare reasonably in terms of marital status, employment, and education, despite the impact therapy had on the development of chronic illnesses and general health. However, differences exist when compared to siblings and additional support services may allow for improved integration.
When interpreting the results of this study, there are some limitations that must be considered. The majority of the outcomes were based upon self-report and thus may be subject to over- or under-reporting. Exceptions to the self-report included the occurrence of second malignancies, where pathology reports were obtained to verify and classify cases. Causes of death were determined using death certificate information. Assessments of chronic health conditions were restricted to individuals over the age 18 years at the time of enrollment in CCSS. While the vast majority of osteosarcoma patients were over 18, our results can not be applied to younger survivors. Lastly, we used the CCSS sibling cohort to compare with the osteosarcoma survivors for a number of outcomes. Since siblings of pediatric cancer survivors may experience a variety of stressors resulting from the psychosocial and familial impact of having a sibling with cancer, the magnitude of risk observed may be an underestimate. Although this group of survivors was treated differently in terms of surgery (more amputations), more varied chemotherapy regimens and supportive care compared to current osteosarcoma patients, this serves as a baseline for future comparisons.
Overall, the childhood osteosarcoma survivors in this cohort did relatively well considering their extensive treatment: however, they are an at-risk population and warrant life-long follow-up for SMNs, adverse medical conditions and issues related to general health and disability/function and pain. Prospective evaluation of current patients with osteosarcoma will be important to assess both acute and long term effects of current osteosarcoma therapy and their impact on survivorship.
Supported by a grant from the National Cancer Institute (U24 CA55727, L. L. Robison, Principal Investigator) and funding to St. Jude Children’s Research Hospital from the American Lebanese Syrian Associated Charities (ALSAC).
No financial disclosures from authors.
Note: The Childhood Cancer Survivor Study (CCSS) is a collaborative, multi-institutional project, funded as a resource by the National Cancer Institute. For information on CCSS investigators and how to access and utilize the CCSS resource, visit www.stjude.org/ccss.