Osteosarcoma in very young children

Experience of the Cooperative Osteosarcoma Study Group


  • Presented in part at the 43rd Annual Meeting of the American Society of Clinical Oncology June 1-5, 2007, Chicago, Illinois.



This study was conducted to investigate presentation, treatment, and outcome in very young children with osteosarcoma.


The authors retrospectively analyzed the data of 2706 consecutive COSS patients with newly diagnosed osteosarcoma and identified 28 (1.0%) patients aged younger than 5 years at diagnosis. Demographic, diagnostic, tumor, treatment-related variables, response, and survival data were analyzed.


Of the 28 preschoolers, 27 presented with high-grade central osteosarcoma of an extremity, and 1 had a secondary osteosarcoma of the orbit. This analysis focused on the 27 patients with extremity tumors. The size of the primary was large (≥one-third of the involved bone) in 20 of 27 patients. Primary metastases were detected in 4 of 27 children. All patients received multiagent chemotherapy, and 11 of 18 analyzed tumors responded well (>90% necrosis) to neoadjuvant chemotherapy. Limb-sparing surgery was performed in 9 cases, ablative procedures were performed in 15, and, in 3 cases, no local surgery was performed. With a median follow-up of 4 years (6.2 years for survivors), 13 patients were alive. Four patients never achieved a complete remission, and 11 developed recurrences; 14 of these 15 patients died. Five-year overall and event-free survival probabilities were 51% (standard error of the mean [SE], 10%) and 48% (SE, 10%). Better survival was correlated with good response to chemotherapy and later time period of diagnosis.


Osteosarcoma is extremely rare in preschool children. These patients often have large tumors that may require mutilating resections. Prognosis is in the range of that reported for older patients. Cancer 2010. © 2010 American Cancer Society.

Osteosarcoma is the most common, yet still rare, primary malignant bone tumor in childhood and adolescence.1 Incidence significantly varies with age. The Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute (NCI) provides the following osteosarcoma incidence data: 5 to 9 years, 2.4 cases per million per year; 10 to 14 years, 7.6 cases per million per year; and 15 to 19 years, 8.2 cases per million per year. Of note, osteosarcoma is extremely rare before 5 years of age with an estimated incidence of .4 cases per million per year.1, 2 Therefore, the information available on very young osteosarcoma patients is mainly restricted to case reports and small series. A review of the literature yielded only a total of 27 well described preschoolers with high-grade osteosarcoma.3-17

To enhance knowledge on the presentation, treatment, and outcome in very young children with high-grade osteosarcoma, we here detail the Cooperative Osteosarcoma Study Group (COSS) experience on 27 children aged younger than 5 years at diagnosis.


We conducted a retrospective investigation on a cohort of 2706 patients from Germany, Austria, and Switzerland with a first diagnosis of high-grade osteosarcoma, who were consecutively enrolled into the COSS registry between 1977 and 2005. Details of their treatment protocols have been reviewed previously.18–22 All studies were accepted by the appropriate ethics and/or protocol review committee(s). Before initiation of therapy, informed consent was obtained from all patients, parents, or legal guardians, as appropriate.

Diagnostic workup and follow-up investigations for primary tumor and metastases varied over time and were performed as previously described.22 Size of the primary tumor was estimated radiographically as follows: tumors measuring less than one-third of the length of the involved bone were defined as being small, all others as large.

Data on patient demographics, tumor characteristics, front-line therapy, and follow-up information were collected prospectively and coded as previously described.22 All patients aged younger than 5 years at diagnosis were reviewed twice by one of the authors (L. K.), and details on these patients—including presentation, treatments, and outcome—were collected from the COSS database and, retrospectively, from status report forms, radiology, pathology and surgery reports, progress letters, and telephone notes available at the data center.

Polychemotherapy was given to all patients according to the treatment protocol active at time of enrollment. Patients who received treatment according to the first COSS trial (ie, COSS-77) received adjuvant chemotherapy only, whereas patients in subsequent COSS trials received neoadjuvant chemotherapy.18-22 All surgical specimens were analyzed histologically. Response to preoperative chemotherapy was assessed and classified according to Salzer-Kuntschik criteria;23 a tumor cell destruction of >90% (Salzer-Kuntschik grades 1-3) was defined as good response.

Definitive surgery time points varied among protocols: before chemotherapy in COSS-77, between weeks 9 and 18 in COSS-80, and between weeks 9 and 11 in the subsequent COSS trials.18-22 Although the type of surgery was not specified, complete removal of the tumor(s) with wide or radical surgical margins should have been attempted; and the COSS study center was available for guidance. The type of surgery was classified as ablative surgery (amputation and rotationplasty) and limb salvage surgery (endoprosthetic replacement and resection with bone auto-, or allografting). Complete surgical remission was assumed only when all detectable tumors were removed during first-line therapy; and if removal was macroscopically complete. The surgical margins were assessed after surgery and classified as radical, wide, marginal, or intralesional. Nonstandard treatment (eg, radiotherapy) was given to selected patients and chosen at the discretion of the participating institutions.

All patients were evaluated on an intent-to-treat basis. Survival was calculated using the Kaplan-Meier method together with standard errors. Overall survival was calculated from diagnosis until death from any cause, and event-free survival until relapse or death, whichever occurred first. Differences between survival curves were evaluated using the log-rank test.24 Patients who never achieved complete surgical remission were assumed to have suffered an event on day one. All statistical analyses were performed in R version 2.1.1 software (http://www.r-project.org).


Patient Characteristics

All of the eligible 2706 patients had a biopsy-proven first diagnosis of high-grade osteosarcoma. A total of 28 (1%) patients aged younger than 5 years at diagnosis were identified. Of the 28 (male, 16; female, 12) identified preschoolers, 27 presented with high-grade central (medullary) osteosarcoma of an extremity and 1 with a secondary osteosarcoma of the orbit. The patient with the secondary osteosarcoma was excluded from further analyses. From the 27 preschoolers, 20 were actual study participants; the remaining 7 were registered in the COSS database, treated according to the same general guidelines, and followed prospectively. Ten of 27 patients were diagnosed before 1990, 7 between 1990 and 2000, and 10 since 2000. Their median age at diagnostic biopsy was 3.9 (range, 2.2 to 4.9) years. The duration of symptoms (pain, swelling, and/or limping) before diagnostic biopsy was reported for 24 of 27 patients. Median duration of symptoms was 4 (range, 1 to 12) weeks. Nine patients had a pathologic fracture (humerus, 4 of 10; femur, 4 of 12; tibia, 1 of 5). More details are provided in Table 1.

Table 1. Patient and Tumor Characteristics of 27 Preschoolers With High-Grade Osteosarcoma of the Extremity
PatientYear of DiagnosisSexAge, yDuration of Symptoms, wkPrimary TumorPathologic FractureHistologyMetastases Site, No.
  • F indicates female; L, large; M, male; OS, osteosarcoma; S, small; wk, week(s); y, years.

  • a

    The size of the primary tumor of the extremities was estimated as follows: tumors measuring less than one-third of the length of the involved bone were defined as being small; all others as large.

11979M4.8Not knownTibia proxLNoTeleangiectatic OS
21982F4.92Femur distalLNoOS, not further specified
31983M4.65Humerus proxLNoFibroblastic OS
41984M2.64Humerus proxLNoOsteoblastic OS
51984M3.83Humerus proxLYesTeleangiectatic OS
61986M4.8Not knownFemur distalSYesTeleangiectatic OS
71987M2.28Femur distalLNoOsteoblastic OS
81988M4.62Humerus proxLNoOS, no predominant pattern
91988F4.74Femur distalLNoTeleangiectatic OS
101989F3.8Not knownFemur proxLYesOsteoblastic OSLung, 2-5
111990F4.3Not knownTibia proxSNoOS, no predominant pattern
121991F4.82Femur distalLNoOsteoblastic OS
131993M2.22Humerus proxLNoOsteoblastic OSLung, >5
141993M4.63Humerus proxSYesOsteoblastic OS
151996M3.34Humerus proxLYesOsteoblastic OS
161997F3.01Femur distalLNoChondroblastic OS
171999F3.94Tibia proxLNoOsteoblastic OSDistant bone, 1
182000M4.66Femur distalLNoOS, not further specifiedLung, bone, >5
192000M3.05Humerus proxSNot knownOsteoblastic OS
202001M3.65Femur distalLYesOsteoblastic OS
212001F2.52Tibia proxSNoOS, not further specified
222002F3.72Femur distalSNoOsteoblastic OS
232002F4.68Femur distalLNoOsteoblastic OS
242003M3.712Femur proxSYesTeleangiectatic OS
252003F4.81Humerus proxLYesTeleangiectatic OS
262004M4.38Tibia proxLYesOsteoblastic OS
272004M3.41Humerus proxLNoOsteoblastic OS

Primary Tumors and Metastases

The primary tumors were located in the femur (12), the humerus (10), and the tibia (5). Of note, only 7 of the 27 primary tumors were considered to be small (less than one-third of the involved bone), whereas 20 were larger.

In each of these 27 cases, histological assessment was performed by a local pathologist; and, in 25 of 27 patients, by a member of the COSS pathology panel. All 27 tumors were described as high-grade central osteosarcoma, with osteoblastic osteosarcoma the most frequent subtype (occurring in 13). Six patients were reported to have telangiectatic osteosarcoma; 4 of these had pathologic fractures. Further details are provided in Table 1.

Primary metastases were detected in 4 of 27 patients (lungs, 2; solitary bone, 1; lung and bones, 1).

Front-line Treatment and Response to Chemotherapy

Details are provided in Figure 1 and Table 2. All patients received chemotherapy, which was initiated at a median of 7 (range, 1 to 43) days after diagnostic biopsy, according to the COSS protocol active at time of enrollment.18-22 Twenty-five patients received local therapy for the primary tumor (surgery, 24; radiotherapy, 1). In 2 children no local therapy was performed because of progressive multifocal disease during chemotherapy. Surgery at the primary site was performed after neoadjuvant chemotherapy in 22 children. Two patients had up front surgery and received adjuvant chemotherapy only.

Figure 1.

Treatment, response to chemotherapy, and surgical remission status at the primary tumor in 27 preschoolers with osteosarcoma is shown.1 The 2 patients had progressive primary metastatic disease during chemotherapy.2 In 1 of the 24 patients, no macroscopically complete remission (CR) was achieved at the metastatic site.3 In 2 of the 21 patients, no microscopically CR was achieved at the metastatic site, leaving a total of 19 of 27 patients with microscopically complete surgical remission at all tumor sites.

Table 2. Treatment and Outcome of 27 Preschoolers With High-Grade Osteosarcoma of the Extremity
Patient NoFront-Line SurgeryPrimary Tumor MetastasesFront-Line ChemotherapyResponseaRelapse Site(s) (Years, Treatment)Outcome
  • Time points of surgery, relapse, and follow-up; recorded as time since diagnosis of osteosarcoma.

  • A indicates doxorubicin; BCD, bleomycin plus cyclophosphamide plus actinomycin-D; CE, carboplatin plus VP-16; CR, complete resection; C, carboplatin; DOD, died of disease; E, VP-16; Gem, Gemcitabine; I, Ifosfamide; IE, Ifosfamide plus VP-16; LCR1, lost for follow-up in first complete remission; M, high-dose methotrexate; NED, no evidence of disease; P, Cisplatin; TroE, Trofosfamid plus VP-16; V, vincristine.

  • a

    Response to neoadjuvant chemotherapy; good response defined as <10% viable tumor.

1Amputation, wide, 2 wksNo dataPrimary surgeryNo data (1.1, no data)DOD, 1.7
2Resection, marginal, 13 wksA, M, BCDNo dataLocal (0.6, CR), lung (0.9, palliation)DOD, 1.1
3Resection, marginal, 11 wksA, M, BCD, VPoorLocal+lung (0.8, palliation)DOD, 1.3
4Amputation, radical, 7 wksA, M, BCDPoorLung (0.5, palliation)DOD, 0.7
5Amputation, radical, 2 wksA, M, BCDPrimary surgeryBone (14.3, CR+A, M, P, I, CE), bone (23, CR)NED, 23.6
6No surgery, radiotherapy, 40 GyA, M, BCDNo surgeryProgressive disease during therapyDOD, 1.1
7Rotationplasty, radical, 12 wksA, M, BCDGoodLung (1.3, incomplete resection+CE)DOD, 1.5
8Resection, radical, 19 wksA, M, P, I, V, ELate surgeryLung+ZNS (4, palliation)DOD, 4.3
9Amputation, radical, 9 wksA, M, P, IGoodLCR1, 13.5
10No surgeryNo surgeryA, M, P, INo surgeryProgressive disease during therapyDOD, 0.9
11Resection, wide, 20 wksA, M, BCD, ILate surgeryNED, 16.4
12Rotationplasty, wide, 14 wksA, M, P, IGoodLCR1, 11.2
13Resection, wide, 23 wksIncomplete resectionA, M, P, ILate surgeryProgressive disease during therapyDOD, 1.4
14Resection, marginal, 15 wksA, M, P, IPoorLocal (0.7, amputation, CR+C, I), Lung+bones (1.1, C, I)DOC, 1.3
15Resection, wide, 10 wksA, M, P, IGoodNED, 10
16Rotationplasty, wide, 13 wksA, M, P, IPoorLung (1.2, CR+CE); lung (3.9, CR), bone (5.3, CR); lung+mediastinum (6.6, Gem)DOD, 7.7
17Amputation, radical, 12 wksMarginal resectionA, M, P, IPoorLung+bones (1.1, TroE)DOD, 1.8
18No surgeryNo surgeryA, MNo surgeryProgressive disease during therapyDOD, 0.2
19Resection, wide, 13 wksA, M, P, IGoodNED, 6.7
20Rotationplasty, wide, 15 wksA, M, P, IGoodNED, 5.9
21Amputation, wide, 12 wksA, M, P, IGoodNED, 6.2
22Rotationplasty, radical, 13 wksA, M, P, IGoodNED, 6.2
23Rotationplasty, wide, 13 wksA, M, P, I, CEGoodLung (0.9, incomplete resection+BCD, M, Interferon)DOD, 1.6
24Rotationplasty, wide, 15 wksA, M, P, IGoodNED, 5.1
25Resection, wide, 15 wksA, M, P, IGoodNED, 4.7
26Amputation, radical, 15 wksA, M, P, IGoodNED, 4.0
27Amputation, wide, 13 wksA, M, P, IPoorNED, 3.5

Limb-salvage procedures were performed in 9 children (proximal humerus, 7; distal femur, 1; proximal tibia, 1). Marginal resections were reported for 3 of these 9 patients; the other 6 underwent either wide or radical resections. Reconstruction with an expandable endoprosthesis was performed in only one patient, a 4.3-year-old girl with an osteosarcoma of the proximal tibia. However, during follow-up, repeated local surgery was necessary 6 times in this child. The humerus was reconstructed by the claviculaprohumero procedure in 1 patient and a titanium Palacos spacer in 2, whereas resection-replantation operations (such as the Tikhoff-Linberg procedure) were used in the remaining 4 children.

Ablative procedures with wide or radical resections were performed in 15 patients. Among the 8 patients who underwent amputations, 4 had a primary tumor in the proximal tibia (knee joint disarticulation, 2; transfemoral amputation, 2), 3 had a primary in the proximal humerus (shoulder disarticulation, 2; forequarter amputation, 1), and 1 patient had a distal femur primary (transfemoral amputation). Rotationplasties were performed in 6 patients with osteosarcoma of the distal femur and, in 1 patient, osteosarcoma of the proximal femur. In the latter child a Type-B-IIIA rotationplasty was performed.25

In 3 patients no surgery at the primary was performed. A macroscopically complete surgical remission (CSR) at the primary tumor site was achieved in 24/27 children at a median of 13 weeks from diagnosis (range, 2 to 23 weeks). In 3 of the 24 patients the resections were marginal after limb salvage procedures.

Of the 4 patients with primary metastatic disease, 2 had no surgery of metastases. In 1 child no CSR of metastases was achieved. In the other case a macroscopic CSR of a solitary, distant bone metastasis was achieved. However, the resection of the solitary bone metastasis was only marginal, and the child subsequently received local radiotherapy with 55.8 Gy. Overall, surgical attempts to remove all tumors to the point of microscopically complete were successful in only 19 of 27 patients.

Data on histological response of the primary to neoadjuvant chemotherapy were available for 18 of 27 patients. Of these, 11 responded well and 7 poorly.

Overall and Event-Free Survival

The median follow-up for all 27 patients was 4.0 (range, .2 to 23.6) years and 6.2 (range, 3.5 to 23) years for the 13 survivors. All survivors were reported to be in CSR (first remission, 12; third remission, 1) at the time of this analysis. Two of them lost to follow-up in first remission, 13.5 and 11.2 years from diagnosis. Actuarial survival at 5 years was 51% (standard error [SE], 10%) (Fig. 2). Thirteen of the 14 deceased patients died from osteosarcoma at a median of 1.3 years (range, 3 months to 7.7 years) from diagnosis, and 1 patient died with active metastatic disease from treatment-related sepsis.

Figure 2.

Kaplan-Meier curve of overall survival is shown for 27 children younger than 5 years of age with osteosarcoma. CI indicates confidence interval; SE, standard error.

The event-free survival rate at 5 years was 48% (SE, 10%). Events as defined occurred in 15 patients: 4 failed to achieve a CSR, and 11 developed a relapse at a median of 1.1 years (range, 6 months to 14.3 years) from diagnosis. The 3 patients in whom surgical resections at the primary tumor were marginal developed local recurrences. One of the 15 patients is alive in third CSR, 23.6 years from initial diagnosis; all others died. Details on relapse treatment and course of disease are provided in Table 2.

There was no correlation between patient's age or gender, duration of symptoms, presence or absence of pathologic fractures, size and localization of the primary tumor, and outcome (Table 3). However, outlook significantly improved in patients who were treated since 2000 (P = .04) and in patients whose primary tumors showed a good response to neoadjuvant chemotherapy (P = .01, Fig. 3). There was no difference in outcome among patients who underwent amputations, rotationplasties, or limb salvage surgery (Table 3). Four of 9 preschoolers who underwent limb salvage surgery remained in first CSR, 4.7 to 16.4 years from diagnosis. Nine of 15 preschoolers who underwent ablative surgery were reported to be alive in first CSR (8 patients) and third CSR (1 patient), a median of 6.2 (range, 3.5 to 23.6) years from diagnosis.

Figure 3.

Kaplan-Meier curves of overall survival is shown for patients with good versus poor response of the primary tumor to neoadjuvant chemotherapy (P = .02, log-rank test).

Table 3. Univariate Analysis of Overall and Event-Free Survival
 No. of Patients5-year OS ± SE, %P5-Year EFS ± SE. %P
  • Abbreviations: EFS, event-free survival; OS, overall survival.

  • a

    In 3 patients, no information was available.

  • b

    In 1 patient, no information was available.

  • c

    In 3 patients, no surgery was performed, and all 3 died from disease.

  • d

    Assessed in the resected primary tumors of 18 of 27 patients; 2 patients had primary surgery, 3 patients had late surgery (ie, <18 weeks from diagnosis), 3 patients had no surgery, and in 1 patient, no data were available.

 Male1642 ± 13 43 ± 13 
 Female1164 ± 15.4055 ± 15.40
Age, y
 <3425 ± 22 25 ± 22 
 ≥3 to <41080 ± 13 70 ± 15 
 ≥4 to <51337 ± 14.1439 ± 13.20
Diagnosis of osteosarcoma
 <19901020 ± 13 20 ± 13 
 ≥1990 to <2000757 ± 19 43 ± 19 
 ≥20001080 ± 13.0480 ± 13.04
Duration of symptomsa
 <5 wks1560 ± 13 47 ± 13 
 ≥5 wks933 ± 16.1633 ± 16.50
Pathological fractureb
 Yes967 ± 16 67 ± 16 
 No1740 ± 12.2834 ± 12.39
Localization of primary tumor
 Femur1250 ± 14 42 ± 14 
 Humerus1048 ± 16 48 ± 16 
 Tibia560 ± 22.7160 ± 22.58
Size of primary tumor
 Small (<one-third)771 ± 17 71 ± 17 
 Large (≥one-third)2044 ± 11.2739 ± 11.16
Primary metastases
 No2360 ± 10 56 ± 10 
 Yes4All deceasedAll deceased
Front line surgical optionc
 Amputation863 ± 17 63 ± 17 
 Rotationplasty771 ± 17 57 ± 19 
 Limb salvage944 ± 17.6844 ± 17.84
Initial complete surgical remissionc
 <15 wks from diagnosis1861 ± 15 56 ± 12 
 ≥15 wks650 ± 20.8350 ± 20.93
Response to neoadjuvant chemotherapyd
 Good, <10% viable tumor1182 ± 12 82 ± 12 
 Poor, ≥10%743 ± 19.0229 ± 18.01


With this study, we provide pertinent information about presentation and outcome derived from a reasonably large cohort of children aged younger than 5 years at diagnosis of high-grade osteosarcoma. The recruitment period spanned 28 years. Given an almost population-based recruitment (providing a total cohort of 2706 patients with a first diagnosis of high-grade osteosarcoma), the low number of identified patients (28 children representing 1% of all osteosarcoma patients—only 1 per year in a population of almost 100,000,000 inhabitants) confirms that osteosarcoma is extremely rare in this age group. These data are in-line with data from the SEER program of the NCI, where the age-adjusted estimated incidence was 0.4 cases per million children per year.2 A review of the literature yielded only a total of 27 well described preschoolers with high-grade osteosarcoma.3

Among the cohort of 28 children, a 4.3-year-old girl with a secondary osteosarcoma of the orbit (and a history of surgery and radiotherapy for bilateral RB) was identified. It is noteworthy that secondary osteosarcoma can occur even in such young patients. This is one of the youngest patients described with secondary osteosarcoma. Germ-line mutations of the RB gene predispose a patient to osteosarcoma, and the latency period between RB and osteosarcoma onset is significantly shorter inside radiation fields.26 Unfortunately, no information on genetic analyses of the RB gene was available on this child, but the bilateral manifestation of RB suggests the presence of a germ-line RB mutation.

In this group of very young patients, all primary osteosarcomas typically arose in the extremities, with a predilection for the metaphyses of long bones. However, the number of primaries affecting the proximal humerus was higher than expected for osteosarcoma in general.22 This observation strengthens preliminary findings from earlier investigations on smaller cohorts.3 When combining the 27 osteosarcoma cases in preschoolers (from the literature) with the current study's cohort of 27 patients, 21 (39%) had a primary in the proximal humerus, 19 (35%) had a primary in the distal femur, 5 (9%) each had a primary in the proximal tibia or proximal femur, and 1 (2%) patient each had a primary in the distal tibia, second metacarpal, T10 vertebra, or orbit.3-17 These findings are congruent with the old hypothesis that regions of highest skeletal growth are particularly vulnerable to development osteosarcoma, even at a very young age.27

Of note, three-quarters of the extremity tumors were large (greater than or equal to one-third) according to a radiographic measure which analyzes size in relation to the length of the affected bone. This compares to an incidence of one-third large tumors with patients of all ages.22 Although osteosarcomas in preschoolers are always unexpected, the high frequency of large tumors cannot be explained by a longer interval between first symptoms and diagnosis in very young versus older patients. Quite to the contrary, lag time in preschoolers was shorter (ie, median of 4 weeks) than the median lag time of 10 to 15 weeks reported for children with osteosarcoma in general.22, 28 During this long-term, multicenter-multinational study imaging techniques have considerably improved. Clearly, more refined methods to measure and analyze tumor volume are necessary to further validate and investigate the observed phenomenon of a high rate of large tumors in very young children. The rate of preschoolers presenting with primary metastases was about the same as that of older patients.29

Overall, the distribution of histologic subtypes was comparable between the study cohort and older patients with osteosarcoma.30 However, in keeping with other authors' results,3 the telangiectatic subtype (22%) was over-represented when compared with older patients, in which the telangiectatic subtype represents less than 10%.31 The high rate of telangiectatic osteosarcoma may explain, in part, the high rate of pathologic fractures (35%) in the study cohort. Indeed, 4 of the series of 6 children with teleangiectatic osteosarcoma experienced pathologic fractures.

In recent years, there has been a successful shift from amputations toward limb-salvage surgery in neoadjuvant osteosarcoma trials.22 In the very young patients described here, however, ablative surgery was still used far more often than limb-salvage surgery, to the point most children lost a major part of the affected limb. The high rate of ablative surgery may partly be explained by the high percentages of large tumors and pathologic fractures; however, lack of suitable, extendable, endoprosthetic replacement for very small bones probably accounted for most of this effect. Our data match the results from previous reports, in which more than two-thirds of the described preschoolers underwent ablative surgical procedures.3-17 In the current study's cohort, rotationplasties were often performed to treat lower extremity tumors to avoid the unsatisfactory function associated with amputations. In contrast to the situation in the legs, it was often attempted to salvage affected arms. Clearly, the use of limb-salvage approaches to treat these skeletally immature patients remains a major challenge. It must be emphasized that, whatever reconstructive technique is used, complete surgery with wide margins is essential.

The actuarial survival rate at 5 years observed in this study's cohort was only 51%, somewhat below the 67% we had previously observed for extremity osteosarcoma in general.22 It must be kept in mind, however, that microscopically complete surgery of all tumor sites was achieved in only 19 of 27 of our patients during front-line therapy.

Local failure of osteosarcoma carries a very poor prognosis.32 The local relapse rate for those children in our cohort who were treated by limb-salvage surgery was unsatisfactory, with 3 local relapses among only 9 patients. However, none of the local relapses in preschoolers occurred after resections with wide margins: all 3 occurred after marginal resections, and all three affected patients subsequently died. Inadequate resection margins have been shown to be the strongest predictive factor for local treatment failure in osteosarcoma, and wide margins must be achieved at all costs.32 The same rigid criteria that apply to osteosarcoma patients in general must also be applied to preschoolers with osteosarcoma. As for older osteosarcoma patients, relapse—either local or systemic—carries a very poor prognosis in preschoolers.33, 34 The rate of preschoolers presenting with primary metastases was about the same as that of older patients.

Besides complete surgical remission, a good histologic response of the primary tumor, assessed after preoperative chemotherapy, has been reported as a key prognostic factor in osteosarcoma.22 This finding is herein confirmed for the subgroup of preschoolers.

In this study, cohort prognosis seemed better for preschoolers who were treated during the last decade, than for those treated earlier. However, long-term follow-up is needed to robustly confirm this trend. Of note, 1 patient was diagnosed with a very late, solitary bone relapse (Table 2). It may be speculated whether this represents a true relapse or rather a second primary osteosarcoma. The long latency period of 14 years, the different histologic subtypes (telangiectatic, initially; osteoblastic, later), and the lesion's site (distal femur, which is the most common site for primary osteosarcoma in puberty) argue for the latter.

The authors once again state that osteosarcoma is exceedingly rare in preschool children. These very young patients often have large, aggressive tumors requiring mutilating resections, especially in the lower limb. Strict adherence to guidelines for chemotherapy and surgical resection, with an emphasis on tumor control and avoidance of inappropriate attempts at limb preservation, is an essential prerequisite to achieve cure. As for all age groups, prospective, cooperative and, in this case, surely multinational trials are essential to learn more about osteosarcoma in the very young.


The authors made no disclosures.


The authors gratefully acknowledge the outstanding support of Ulrike Pötschger, MSc, and Mathias Kevric in statistic analyses and data management.