Second solid malignancies among children, adolescents, and young adults diagnosed with malignant bone tumors after 1976

Follow-up of a children's oncology group cohort




The growing number of individuals surviving childhood cancer has increased the awareness of adverse long-term sequelae. One of the most worrisome complications after cancer therapy is the development of second malignant neoplasms (SMNs).


The authors describe the incidence of solid organ SMN in survivors of pediatric malignant bone tumors who were treated on legacy Children's Cancer Group/Pediatric Oncology Group protocols from 1976 to 2005. This retrospective cohort study included 2842 patients: 1686 who were treated for osteosarcoma (OS) and 1156 who were treated for Ewing sarcoma (ES).


The cohort included 56% boys/young men and 44% girls/young women, and the median age at primary diagnosis was 13 years. The median length of follow-up was 6.1 years (range, 0-20.9 years). In this analysis, 64% of patients were alive. Seventeen patients with solid organ SMN were identified. The standardized incidence ratio was 2.9 (95% confidence interval [CI], 1.4-5.4) for patients who were treated for OS and 5.0 (95% CI, 2.6-9.4) for patients who were treated for ES. The median time from diagnosis to development of solid SMN was 7 years (range, 1-13 years). The 10-year cumulative incidence of solid organ SMN for the entire cohort was 1.4% (95%CI 0.6%-2%).


The magnitude of risk of solid SMNs was modest after treatment for malignant bone tumors. However, radiation-related solid SMNs will increase with longer follow-up. Because nearly 33% of patients die from their disease, recurrence remains the most significant problem. The development of improved therapies with fewer long-term consequences is paramount. Follow-up should focus on monitoring for both recurrence of primary malignancies and development of SMNs. Cancer 2008. © 2008 American Cancer Society.

The 2 most common malignant bone tumors in children and adolescents are osteosarcoma and Ewing sarcoma family of tumors (ESFT). According to SEER, from 1975 to 1995,1 the annual incidence rate of malignant bone tumors was 8.7 per million children aged <20 years, and approximately 400 children are diagnosed with osteosarcoma and 200 children are diagnosed with ESFT each year in the United States. The 5-year survival is improving for children with malignant bone tumors and, with aggressive, multimodal therapy, nearly 70% are expected to be long-term survivors. One of the most worrisome complications after cancer therapy is second malignant neoplasms (SMNs). The overall cumulative incidence of SMN in survivors of childhood cancer is reported as 3.2% at 20 years but varies from 1.6% to 7.6%, depending on the primary diagnosis.2

Because osteosarcoma generally is considered radioresistant, chemotherapy and surgery are the primary therapeutic modalities for this disease. The backbone of osteosarcoma chemotherapy includes cisplatinum, doxorubicin, and high-dose methotrexate. The addition of other agents, such as carboplatinum, ifosfamide, and etoposide, have been investigated.3–5 Although osteosarcoma usually is sporadic, a rare patient may have an underlying cancer predisposition syndrome, namely, Li-Fraumeni syndrome, possibly resulting in an increased risk of SMNs because of genetic susceptibility. There is limited information regarding the risk of SMNs after treatment for osteosarcoma.6, 7

Therapy for ESFT involves multiagent chemotherapy and local control with either surgery and/or radiation. During the 1980s and 1990s, the backbone of therapy included vincristine, doxorubicin, and cyclophosphamide. Therapeutic trials have demonstrated significant survival advantage by the addition of ifosfamide and etoposide for patients with localized disease.8, 9 In contrast to osteosarcoma, Li-Fraumeni typically is not associated with ESFT.

Several studies have examined the risk of SMN after treatment for ESFTs. However, small sample size,10 varying therapies,11 short follow-up,12 and single-institution studies13 limit the generalizability of those results. Although most previous reports have focused on secondary leukemia or myelodysplasia occurring after bone tumors, only a few have focused on solid tumors developing in this patient population.14 The objectives of this study were to determine the incidence of solid SMNs and to identify potential risk factors in children and adolescents with malignant bone tumors entered on legacy Children's Cancer Group (CCG)/Pediatric Oncology Group (POG) protocols since 1976.


CCG/POG conducted clinical trials in cooperation with member institutions throughout the United States, Canada, Australia, and selected countries in Europe. The member institutions were required to register all newly diagnosed cancer patients with the operations office; then, eligible patients were entered into active therapeutic protocols. The operations office was responsible for determining patient eligibility, randomized assignments to the appropriate therapeutic arms (if necessary), and follow-up of patients for all potential outcomes. Member institutions were required to submit semiannual follow-up reports on all patients enrolled in therapeutic protocols. Those reports included information on survival status, disease status, and development of second malignancies for all patients.

Our cohort consisted of 2842 patients with newly diagnosed osteosarcoma or ESFT aged <33 years at diagnosis who were enrolled in 1 of the 11 therapeutic protocols for untreated osteosarcoma or ESFT conducted between 1976 and 2002. The distribution of patients according to the therapeutic trials is summarized in Table 1. Informed consent was obtained from patients, parents, or guardians at the time of enrollment. Clinical results of many of the trials, with the therapeutic plans, have been published (see the references indicated in Table 1).

Table 1. Treatment Protocols, Timing, and Distribution of Patients Included in This Cohort Study
ProtocolReferenceDiagnosisDate OpenedDate ClosedNo. of Patients (%)*Agents Used
  • OS indicates osteosarcoma; CCG, Children's Cancer Group; POG, Pediatric Oncology Group; ESFT, Ewing sarcoma family of tumors.

  • *

    Not all patients received each agent on the protocol because some protocols were varying pilot studies, and some were randomized clinical trials.

  • Chemotherapy agents: D indicates doxorubicin; M, methotrexate; V, vincristine; P, cisplatin, A, actinomycin; B, bleomycin; C, cyclophosphamide; I, ifosfamide, Ca, carboplatin; E, etoposide.

CCG741Krailo 198725OSSept 1976Oct 1981257 (9%)D,M,V
CCG782Provisor 199726OSAug 1983Oct 1986232 (8%)P,D,M,A,V,B,C
CCG7921Meyers 200527OSOct 1993Nov 1997777 (27%)P,D,M,I
POG8651Goorin 200328OSMar 1987Mar 1993100 (4%)P,D,M,A,B
POG9259Goorin 199829OSJan 1992Nov 199437 (1%)P,D,M,I,Ca
POG9450Goorin 20024OSMay 1995Sept 199741 (1%)P,D,M,I,E
POG9754Schwartz 200430OSSept 1999Feb 2002242 (9%)P,D,M,I,E
CCG7881Grier 20039ESFTDec 1988Nov 1992324 (11%)D,I,E,V,A,C
CCG7942Granowetter 200131ESFTMay 1995Sept 1998465 (16%)D,I,E,V,C
POG8850Grier 20039ESFTDec 1988Dec 1994257 (9%)D,I,E,V,A,C
POG9457Bernstein 200632ESFTAug 1999Mar 2000110 (4%)D,I,E,V,C

For each protocol, a therapeutic summary was prepared that included the dose of radiation therapy (and assigned fields) and chemotherapeutic exposures. Assigned cumulative doses were calculated for cisplatinum, doxorubicin, etoposide, cyclophosphamide, and ifosfamide. The use of chemotherapy and radiation therapy was categorized as dichotomized variables (yes/no) and was included in the analysis using the intent-to-treat method. The total cumulative doses ranged from 0 to 480 mg/m2 of body surface area for cisplatinum, from 90 to 600 mg/m2 for doxorubicin, from 0 to 5000 mg/m2 for etoposide, from 0 to 21.6 g/m2 for cyclophosphamide, and from 0 to 140 g/m2 for ifosfamide. Assigned radiation doses depended on tumor type, protocol, and site of disease.

For patients who had second neoplasms, the date of diagnosis, histologic characteristics, and tumor site were recorded. Pathology reports were requested from treating institutions and were reviewed to verify diagnoses.

The time at risk for second neoplasms was computed from the date of diagnosis of osteosarcoma/ESFT to the date of diagnosis of second neoplasm, date of death, or date of last contact, whichever came first. The end of follow-up for the study was April 1, 2005. Overall and event-free survival were calculated using actuarial methods. Cumulative incidence of second solid malignancy (overall and by histologic subtypes) over time was calculated and compared across characteristics of interest by using the method proposed by Gray.15 To estimate the risk of second solid neoplasms, the numbers of person-years of observation were complied for subgroups of the cohort that were defined by age and sex. Rates of incidence of cancer (obtained from the registry of the Surveillance, Epidemiology, and End Results [SEER] Program of the National Institute of Health)16 were used to calculate the expected number of cases of cancer. The standardized incidence ratio (SIR) was calculated as the ratio of observed to expected cases. The 95% confidence interval (CI) for the SIR was calculated assuming that the number of cases followed a Poisson distribution.17


The key characteristics of the cohort are summarized in Table 2. The median age at diagnosis was 13 years (range, 0-33 years) for patients with malignant bone tumors. The median follow-up from the date of enrollment to the date of last contact was 6.1 years (range, 0-20.9 years) for the 1812 patients who were alive at last contact. The date of last contact was within 3 years for 46% of patients who were alive before April 1, 2005. The overall survival rate for this cohort was 60% at 10 years (Fig. 1, top), and there was a significant difference in overall survival based on the log-rank test; however, the ultimate cure rate did not differ significantly by primary diagnosis (Fig. 1, bottom). Seventeen patients developed a solid SMN at a median of 7 years (range, 1-13 years) from the primary diagnosis. The histologic subtypes included breast cancer (n = 3 patients); malignant fibrous histiocytoma (n = 3 patients); osteosarcoma (n = 2 patients); and chondrosarcoma, leiomyosarcoma, undifferentiated sarcoma, undifferentiated carcinoma, lung adenocarcinoma, renal cell carcinoma, ovarian cancer, papillary carcinoma and anaplastic astrocytoma (n = 1 patient each) (Table 3).

Figure 1.

Overall survival of all patients with bone tumor (top) and of patients with bone tumor according to primary diagnosis (Ewing sarcoma vs osteosarcoma [OS]) (bottom).

Table 2. Characteristics of the Patient Population With Second Neoplasms
CharacteristicNo. of Patients (%)
Total CohortPatients With Solid Second Neoplasms
All Solid SMNsBreastSecondary Sarcomas
  1. SMNs indicates second malignant neoplasms.

No. of patients28421739
Type of disease    
 Osteosarcoma1686 (59)833
 Ewing sarcoma1156 (41)906
 Boys/men1579 (56)5 (29)0 (0)4 (44)
 Girls/women1263 (44)12 (71)3 (100)5 (56)
Age at diagnosis of primary sarcoma, y    
 Median [range]13 [0-33]14 [2-17]15 [14-17]8 [2-16]
Time from study enrollment to second neoplasm, y    
 Median [range]7 [1-13]7 [5-13]6 [2-12]
Age at diagnosis of second neoplasm, y    
 Median [range]19 [8-31]23 [19-31]16 [8-26]
Therapeutic exposure, yes    
 Cisplatinum1429 (50)6 (35)1 (33)3 (33)
 Doxorubicin2842 (100)17 (100)3 (100)9 (100)
 Etoposide1120 (39)8 (47)0 (0)5 (56)
 Cyclophosphamide1488 (52)15 (88)1 (33)9 (100)
 Ifosfamide1625 (57)8 (47)0 (0)5 (56)
 No1329 (47)5 (29)1 (33)3 (33)
 Yes1513 (53)12 (71)2 (67)6 (67)
 Died1030 (36)7 (41)1 (33)4 (44)
 Alive1812 (64)10 (59)2 (67)5 (56)
Table 3. Key Clinical Information on Patients With Solid Second Malignant Neoplasms
PatientInitial DiagnosisTime to SMN Diagnosis, yTreatment: Yes/NoSolid SMN Cancer Type
Cancer TypeCOG Study No.Age at Primary Diagnosis, yCyclophosphamideEtoposideRadiation
  1. COG indicates Children's Oncology Group; SMN, second malignant neoplasm; L, left; R, right.

1Osteosarcoma741145NoNoYesInfiltrating ductal carcinoma of L breast
2Osteosarcoma7411714NoNoYesInfiltrating colloid mucinous carcinoma L breast
3Osteosarcoma782161YesNoNoAdenocarcinoma of lung L chest
4Osteosarcoma7821412YesNoNoLeiomyosarcoma originating in radiated R orbital field
5Osteosarcoma782162YesNoNoMalignant fibrous histiocytoma of bone (orbit/sinus)
6Osteosarcoma782158YesNoNoBreast cancer
7Osteosarcoma78287YesNoNoMalignant undifferentiated tumor of R chest wall
8Ewing sarcoma78811310YesYesYesOvarian tumor (stage I)
9Ewing sarcoma788179YesYesYesMalignant fibrous histiocytoma of bone (orbit)
10Ewing sarcoma788175YesYesYesChondrosarcoma vs chondroblastic osteosarcoma
11Ewing sarcoma7881158YesYesYesOsteosarcoma
12Osteosarcoma86511410YesNoYesAnaplastic astrocytoma
13Ewing sarcoma8650510YesNoYesOsteosarcoma
14Ewing sarcoma865026YesYesYesMalignant fibrous histiocytoma of bone (femur)
15Ewing sarcoma7942136YesYesYesPoorly differentiated sarcoma with myogenic features
16Ewing sarcoma794278YesYesYesPapillary carcinoma of the thyroid
17Ewing sarcoma7942157YesYesYesRenal cell carcinoma

The estimated cumulative incidence rate for solid SMNs was 1.4% (standard error [SE], 0.4%) at 10 years for the entire cohort (Fig. 2, top), 0.9% (SE, 0.4%) for the osteosarcoma group, and 1.8% (SE, 0.65%) for the ESFT group (Fig. 2, bottom). Compared with a sex- and age-matched general population, patients with ESFT were at a 5-fold increased risk of developing a solid second malignancy (SIR, 5.0; 95% CI, 2.6-9.4), whereas patients with osteosarcoma were at a 2.9-fold increased risk (SIR, 2.9; 95% CI, 1.4-5.4).

Figure 2.

Cumulative incidence of second malignant neoplasms among all patients with bone tumor (top) and among patients with bone tumor according to primary diagnosis (Ewing sarcoma vs osteosarcoma) (bottom).

Certain therapeutic exposures were associated with increased risk of solid SMNs. Exposure to etoposide (SIR, 4.8; 95% CI, 2.5-9.5) and cyclophosphamide (SIR, 5.8; 95% CI, 3.5-9.5) was associated with an increased risk of solid SMNs. Exposure to radiation therapy also increased the risk of solid SMNs (SIR, 4.1; 95% CI, 2.4-7). At a median follow-up of 9.3 years (range, 3-8.7 years) for the 17 patients who developed solid SMNs, the estimated 5-year overall survival rate after patients developed a solid SMN was 38% (Fig. 3).

Figure 3.

Overall survival of 17 patients with solid second malignant neoplasms (SMN) after the date their solid SMN was diagnosed.

Sensitivity Analysis

At the time of the current analysis, contact was documented with 46% of the patients who remained alive within the previous 3 years. To alleviate potential problems related to incomplete follow-up, sensitivity analyses were performed in which all patients who had not had an event (death or SMN) were censored as of April 1, 2005, irrespective of their actual date of last contact. We assumed that those who were lost to follow-up did not develop an SMN. With these extremely conservative assumptions, a lower boundary was placed on the reported estimates. The results of the sensitivity analysis are presented in Table 4.

Table 4. Standardized Incidence Ratio: Sensitivity Analysis
VariableEstimate From Observed DataFollow-Up Until April 1, 2005
  1. SIR indicates standardized incidence ratio; CI, confidence interval; SMN, second malignant neoplasm.

All SMNs3.732.35-5.991.570.99-2.52
By primary diagnosis    
 Ewing sarcoma4.962.64-9.422.791.49-5.30
By type of SMN    
 Breast cancer27.8010.05-
By key therapeutic exposures    


This retrospective cohort study revealed that the cumulative incidence of solid SMNs in patients who are diagnosed with primary malignant bone tumors is relatively low with a cumulative incidence of 1.4% at 10 years. Comparable risks of second solid malignancies were observed in patients with ESFT and osteosarcoma. We observed that radiation, cyclophosphamide, and etoposide may be associated with an increased risk of solid SMNs. The outcome after development of a solid SMN is poor with a 5-year overall survival rate <40%.

The strength of this study lies in the size of the cohort and, despite limitations, the quality of the data. The COG cohort of patients with malignant bone tumors include more patients than are included in the SEER and Childhood Cancer Survivorship Study (CCSS) databases. Of the 752 children who were diagnosed with osteosarcoma and the 479 children who initially were diagnosed with Ewing sarcoma and were followed in the SEER database, there were 10 SMNs and 15 SMNs reported with 15 and 44 excess cancers per 10,000 person-years, respectively.18 In the initial report from CCSS, 25 of 1136 patients who were followed with malignant bone cancer developed an SMN.19 Of these, it is noteworthy that 6 were breast cancers. There was a 17.9 absolute excess risk for second cancers per 10,000 person-years of follow-up. The outcome was not reported separately for osteosarcoma and Ewing sarcoma. Data from the COG cohort expand on and corroborate the findings from these studies.

Earlier studies reported a high incidence of SMNs in patients with ESFT, with the magnitude of risk approaching 30% at 10 to 20 years after treatment.20, 21 More recent studies suggest a lower incidence ranging from 1% at 5 years to 6.5 percent at 10 years.11–13, 22 Although each of these studies was based on relative small sample sizes, most identified radiation as playing a role in the risk of developing a solid SMN. The current study revealed that the cumulative incidence of solid SMNs in patients with ESFT in this cohort was 1.8% at 10 years. We anticipate that the incidence will rise with increasing follow-up, as has been reported for radiation-related second malignancies in other cohorts.

There are fewer reports on the incidence of SMNs in survivors of osteosarcoma. In our study, the overall 10-year cumulative incidence of solid SMNs was 0.9%. Others have reported a 10-year cumulative incidence of 2% to 3% at 10 years, but those studies also included hematologic malignancies.6, 7 Although the relative risk of solid SMNs in long-term survivors of osteosarcoma was higher than in the general population in the current study, the magnitude of risk was lower than that for Ewing sarcoma. This likely reflects increased use of radiation in patients who survive Ewing sarcoma, but it also may relate to higher cumulative doses of alkylating agents and etoposide in patients with Ewing sarcoma (data not shown). It is worth noting that all 3 of the solid SMNs of the breast arose in survivors of osteosarcoma. These, and perhaps others, may be patients with Li-Fraumeni syndrome, which is characterized by the development of sarcomas and early breast cancers and by an exceptionally high risk of developing multiple primary cancers.23

Thus, our study, with relatively large numbers, confirms the reports of others11–13, 22 that the magnitude of risk of solid SMNs is modest after treatment for primary bone cancer. There are 2 major possible explanations for the low cumulative incidence of solid SMNs. First, and more optimistically, there may be fewer solid SMNs in the recent era related to improved therapeutic techniques. For example, it is possible that a patient may be able to achieve adequate local control with surgery alone after neoadjuvant therapy, thus omitting the need for radiation. In addition, radiation techniques have improved substantially, reducing the normal tissue exposure. The second reason for the low cumulative incidence may stem from the finding that the cohort has been followed for a median of 6 years, and that, with increasing follow-up, radiation-related solid malignancies will emerge, because they typically occur after a latency of 10 to 15 years.

However, like in other large cohorts, there is patient attrition with extended follow-up. To compensate for the impact of loss to follow-up on the magnitude of risk, we conducted sensitivity analyses based on the assumption that patients who are lost to follow-up are alive and have not had an adverse event of interest. This extremely conservative assumption allowed us to place a lower limit on the magnitude of risk, thus permitting an estimate of the true risk somewhere between the reported value and that derived by these sensitivity analyses. Another limitation relates to the small numbers of events that preclude any significant examination of the risks for SMNs or any meaningful subset analyses. Also, lack of family history data prevented us from predicting whether any patients were likely to have Li-Fraumeni or another cancer predisposition syndrome. Finally, we were not able to assess accurately whether SMNs arose within the radiation field.

Methods to assure the ongoing, long-standing follow-up of children who survive cancer are necessary. The COG has instituted long-term follow-up guidelines that may help assure ongoing surveillance of late effects.24 Additional resources are necessary and currently are being instituted by the COG to assure the long-term follow-up of cancer survivors.