Primary osteosarcoma of the ribs: A report from the Cooperative Osteosarcoma Study Group
Abstract
Purpose
Primary rib osteosarcoma has not been investigated extensively, and clinical characteristics and optimal therapeutic strategies have not been defined. The authors used the database of the Cooperative Osteosarcoma Study Group (COSS) to analyze this tumor-site in depth.
Methods
The COSS database was searched for treatment-naive, high-grade osteosarcomas of the rib. Affected patients were analyzed for demographic and tumor-related factors, treatments, and outcomes.
Results
A total of 44 patients (23 males, 21 females; median age, 23 years [range, 6–59]) were identified. Primary metastases were detected in six of 44 (14%) patients. Surgery was performed in 40 of 44 (91%) patients and rendered 35 of 44 (80%) patients macroscopically disease-free. Chemotherapy was known to have been administered in 43 of 44 (98%) patients and radiotherapy in seven of 42 (17%) (no data for two patients). A good response to chemotherapy was only noted in five (33%) of those 15 evaluable patients who had received any preoperative chemotherapy. After a median follow-up of 2.49 (0.22–40.35) years for all patients and 6.61 (0.25–40.35) years for 26 survivors (21 of these in first complete remission), 5-year actuarial overall and event-free survival were 53.0% (8.5%) and 42.2% (8.1%), respectively. Incomplete tumor surgery was the most notable negative prognostic factor. Osteoblastic histology and a poor response to chemotherapy may have contributed.
Conclusion
This large series provides evidence that patients with costal primaries are older than the average osteosarcoma patient, but appear to share the similar tumor biology and—if treated according to standard protocols—prognostic factors with tumors of other sites. Early, preoperative diagnosis and permanent, definitive local control remain major challenges and should contribute to improved outcomes.
INTRODUCTION
Osteosarcomas are rare but typical malignancies of children, adolescents, and, less frequently, adults. They usually arise in an extremity and are treated with surgery and multidrug chemotherapy, leading to cure in well over half of all cases. Much less commonly, the disease may also occur in other sites of the body. Presentation in the trunk is often associated with reduced operability and inferior outcomes.2 Presentation in the ribs, however, is extremely rare, thus little is known about the optimal therapy and the ensuing disease course. So far, only case reports and limited case series about this unusual disease site have been published,3, 4 and they provide very limited information. Clinicians caring for affected patients are therefore faced with considerable uncertainties.
The Cooperative Osteosarcoma Study Group (COSS) has been registering and following osteosarcoma patients since 1977, leading to one of the world's largest disease-oriented databases.5 This allows analyses of even rare presentations of osteosarcoma to be performed with reasonable patient numbers. We searched the COSS database for primary, high-grade central osteosarcomas arising in a rib. Affected patients were then analyzed for baseline characteristics, treatment, and outcome to learn more about this rare disease manifestation.
MATERIALS AND METHODS
Patient selection and data collection
The COSS-database of 5443 patients registered from 1980 to 2020 was searched for all primary, previously untreated high-grade central osteosarcomas of bone arising within a rib. The search revealed 65 registered costal primaries. Twenty-one of these had to be excluded for the reasons detailed in Figure 1, leaving 44 patients from 32 participating centers for further analyses.

Consolidated Standards of Reporting Trials diagram. HGS, high-grade surface; LGC, low-grade central.
Details of recruitment and treatment protocols have been described previously.5-7 Registration of all patients with osteosarcoma into the COSS-database, in which over 200 institutions have entered patients, is encouraged for Germany, Austria, and Switzerland. It is assumed to be near complete for affected children and adolescents. We have no means to assess how high this proportion is among eligible adults but must assume that it is far less complete. Data on patient demographics, tumor characteristics, front-line therapy, and follow-up information were collected prospectively and also coded as already described.2 Further information was collected retrospectively from status report forms, radiology, pathology, and surgery reports, as well as progress letters. All COSS studies and registries had been accepted by the appropriate ethics and/or protocol review committees. Informed consent was required from all patients and/or legal guardians, whichever appropriate, before enrollment.
Diagnostic workup and staging procedures
For diagnosis, standard histologic investigation and suitable immunohistochemistry studies were performed according to local practice and, if material was made available, by one of a panel of reference pathologists. Staging procedures prescribed by all COSS protocols included conventional radiography of the tumor, chest radiography, computed tomography of the thorax, and a whole-body 99Tc-methylene-diphosphonate bone scan. Computed tomography, magnetic resonance tomography of the primary site, and positron emission tomography were done according to time and availability.
Treatment
Treatment for patients with costal tumors followed the guidelines for extremity primaries set forth in the various COSS protocols.5-7 All patients were to receive multi-drug, pre-, and postoperative chemotherapy. The drugs used varied with time but almost always included high-dose methotrexate, doxorubicin, and cisplatin, with ifosfamide being prescribed for some. A minority were to receive other drugs.5-7
Local treatment of the primary was to be surgery with wide margins according to Enneking et al.8 whenever feasible, with radiotherapy on an individual basis for patients in whom such margins could not be accomplished. Primary metastases, if present, were also to be removed surgically whenever feasible. Response of the primary tumor to preoperative chemotherapy was graded according to Salzer-Kuntschik et al.,9 with a good response defined as <10% viable cells in the resected specimen. A patient was considered to have achieved a surgically complete remission if all foci were removed at least macroscopically.
Follow-up
Follow-up guidelines varied with time. Generally, the search for potential local recurrences was by conventional x-ray for at least every 3 months for 4–5 years after treatment and only in case of clinical suspicion thereafter. Lung metastases were to be searched for by conventional chest x-rays, recommended every 4–8 weeks during years 1–2, every 8–12 weeks in years 3 and 4, and every 6 months from year 5 until years 8–10. Later follow-up was recommended but left to the treating physician.
Statistical analyses
All hospitals that had reported surviving patients at last follow-up were contacted in writing and asked for current patient status information. All patients were then evaluated on an intent-to-treat basis. Where appropriate, descriptive statistics were used to describe patient, tumor, and treatment characteristics. Survival was calculated using the Kaplan–Meier method,10 together with SEs and reported 2- and 5-year survival estimates. Overall survival was calculated from the date of biopsy until death from any cause, and event-free survival was calculated until relapse or death, whichever occurred first. Patients who never achieved a macroscopically complete surgical remission were assumed to have suffered an event on day one. Information on second primary malignancies was collected separately. The log-rank test was used to compare survival curves.11 Statistical analyses were performed using the SPSS statistical software packet (SPSS Statistics for Windows, version 28.0.1.0; IBM, Armonk, NY).
RESULTS
At diagnosis, the 44 eligible patients were a median of 23 years old (range, 6–59) including four children (≤12), nine adolescents,13-17 and 31 adults (18+). There were 23 (52%) affected males and 21 (42%) females. Tumors arose secondarily to previous malignancies in seven (16%) patients, six of these were adults (≥18 years), all in known or suspected previous radiation fields (two Hodgkin lymphoma, two breast cancer, one renal cancer, one acute lymphoblastic leukemia, and one patient with both pulmonary and gastric carcinomas). A tumor-predisposition syndrome (Li-Fraumeni syndrome) was known in one (2%) patient only.
The median duration of symptoms before diagnostic biopsy in 39 of 44 (89%) patients with available information was 61 (0–1271) days (28, pain; 23, swelling; two, pathological fracture; one, dyspnea caused by lung metastases; and one, paralysis due to spinal ingrowth) (multiple mentions possible). A malignant cause was suspected in 32 of 42 (76%) patients with appropriate information, and a benign cause was supposed in 10 of 42 (24%). The exact tumor location was known for 43 of 44 (98%) patients, in whom 22 primaries arose in the first to sixth ribs, 18 in the seventh to twelfth ribs, and three overlapped (one unknown). A maximal tumor diameter estimate was documented for 34 of 44 (77%) patients (18, x-ray; 13, magnetic resonance imaging; one, computed tomography; one, positron emission tomography; and one ultrasound). For these 34 patients, it was 7 (range: 2–14) cm.
The osteosarcoma subtype was known for 33 of 44 (75%) patients and was considered osteoblastic in 15, chondroblastic in four, fibroblastic in two, teleangiectatic in two, small cell in three, and secondary in seven. Initial staging revealed unequivocal primary metastases in six of 44 (14%) patients (two, lung; one, bone; two, lung and bone; and one, lung, bone, and elsewhere). The other 38 of 44 (86%) patients were considered to have localized disease.
A total of 40 of 44 (91%) patients went on to receive surgery of their primary (three, no surgery and one, lost previously). Of 39 of 40 patients with appropriate information, 28 of 39 (72%) were operated on knowing the correct diagnosis of osteosarcoma and 11 of 39 (28%) were operated on under a different assumption. Surgery could be considered microscopically complete in 23 of 40 (58%) patients, macroscopically complete in 13 of 40 (33%; six of these with documented microscopic tumor-rests) patients, and macroscopically incomplete in four of 40 (10%) patients. Overall, 35 of 44 (80%) patients were considered to have achieved an at least macroscopically complete remission of all tumor sites. Nine of 44 (20%) others retained tumor residuals (incl. 1 patient who developed metastases while still on preoperative chemotherapy).
A total of seven of 42 (17%) patients (two with no data) went on to receive radiotherapy to their tumor. This included one patient each receiving radiotherapy instead of surgery, with macroscopic, or with microscopic tumor rests, and four in complete local surgical remission. The dose was known in five and was a median of 66 (30.7–66) Gy.
A total of 43 of 44 (98%) patients (one of 44 lost before initiation) were known to have received systemic chemotherapy (14 pre- and postoperatively, one preoperatively only, 25 postoperatively only, and three chemotherapy without surgery). Their time period from diagnostic procedure (biopsy or primary surgery) to the initiation of chemotherapy was 26 (5–221) days. The choice to use preoperative chemotherapy did not correlate with tumor size (p > .1, χ2). In 41 of 43 (95%) systemically treated patients with appropriate information, its documented length was 238 (54–591) days. In total, 41 of 43 (95%) patients with available data were known to have received treatment with at least three cytotoxic agents. Forty of these (93%) received at least three of the four drugs doxorubicin, high-dose methotrexate, cisplatin, and ifosfamide (chemotherapy details unknown in the remaining patients). Additional cytotoxic therapy was documented in 14 cases, targeted therapy (sorafenib) in only one. Among the 15 patients known to have received preoperative chemotherapy, there were five (33%) good (<10% viable tumor) and 10 (67%) poor responders.
After a median follow-up of 2.49 (range, 0.22–40.35) for all 44 patients and 6.61 (range, 0.25–40.35) years for survivors, 26 of 44 (59%) patients remained alive (two without having achieved complete remission (CR) [follow-up, 0.441 and 0.616 years], 21 in first remission, one in second remission, one in third remission, and one alive with disease at first recurrence). A total of 18 of 44 (41%) patients had died. Causes of death were osteosarcoma in 15 patients (six without ever having achieved a CR, seven in first, one in second, and one in third recurrence). One patient each died of perioperative complications of attempted metastasectomy, of a secondary malignancy (ovarian cancer as a third tumor while still in first remission of osteosarcoma), and of unknown causes at first recurrence, respectively.
Events as defined occurred in 23 of 44 patients. Nine patients failed to achieve a first CR, nine patients developed a local, two a metastatic, and one a combined recurrence, the site of recurrence was unknown in another. One further patient died of a secondary malignancy while in remission of osteosarcoma (see above). At recurrence, all metastases involved the lungs. Twenty-one of 44 patients remained event-free. The projected 2- and 5-year overall and event-free survival rates for all 44 patients were 65.7% (SE = 7.7%) of 53.0% (SE = 8.5%) and 54.3% (SE = 7.8%) of 42.2% (SE = 8.1%), respectively (Figure 2). Among the evaluated potential risk factors, the following did not correlate with survival estimates (p > .1): patient age at below or of at least 18 years, male or female patient gender, a history or no history of previous malignancy, a duration of pre-diagnostic symptoms of less than 60 days or above, a tumor site in the upper or lower half of the thorax, a tumor size below 7 cm or larger, the presence or absence of primary metastases, a biopsy before 2000 or thereafter, a biopsy with or without suspicion of malignancy, the use of preoperative chemotherapy versus postoperative chemotherapy only, a chemotherapy duration of below or at least 240 days, the use of radiotherapy or none, nor a good or a poor tumor response to preoperative chemotherapy. The osteoblastic subtype trended to do worse than others (p = .056), failure to operate all tumor sites was a significant predictor of inferior survival (p < .001). Only histologic subtype (p = .076) and tumor response to preoperative chemotherapy (p = .077) showed a trend to correlate with the probability of event-free survival. Here, the extent of surgery (complete vs. incomplete) was not evaluable by definition (Table 1).

Overall (blue) and event-free (red) survival of all 44 evaluable patients with an osteosarcoma of the rib.
Patients | No. (%) | Overall survival (%) | p, log-rank | Event-free survival (%) | p, log-rank | ||
---|---|---|---|---|---|---|---|
2-year | 5-year | 2-year | 5-year | ||||
All | 44 (100) | 65.7 (7.7) | 53.0 (8.5) | — | 54.3 (7.8) | 42.2 (8.1) | — |
Age, years | |||||||
<18 | 13 (30) | 66.7 (13.6) | 58.3 (14.2) | .545 | 59.8 (14.0) | 59.8 (14.0) | .267 |
≥18 | 31 (70) | 65.3 (9.4) | 49.9 (10.7) | 52.0 (9.4) | 32.6 (9.8) | ||
Sex | |||||||
Male | 23 (52) | 65.2 (10.6) | 49.6 (11.3) | .895 | 51.6 (11.2) | 41.3 (11.1) | .931 |
Female | 21 (48) | 66.4 (11.3) | 56.9 (13.1) | 56.4 (11.0) | 42.7 (12.1) | ||
Previous malignancy | |||||||
None | 37 (84) | 66.0 (8.4) | 55.6 (8.9) | .751 | 56.8 (8.5) | 46.3 (8.9) | .430 |
Present | 7 (16) | 66.7 (19.2) | 33.3 (25.5) | 42.9 (18.7) | 21.4 (17.8) | ||
Prediagnostic symptoms | |||||||
≤60 days | 19 (49) | 77.0 (10.1) | 70.0 (11.4) | .156 | 62.0 (11.4) | 55.8 (11.8) | .166 |
>60 days | 20 (51) | 55.6 (11.7) | 43.8 (11.9) | 42.9 (11.4) | 31.3 (10.9) | ||
No data 5 | |||||||
Tumor site 5 | |||||||
First or sixth rib or overlapping | 24 (56) | 72.2 (9.7) | 55.1 (11.4) | .991 | 61.4 (10.1) | 44.1 (11.3) | .824 |
Seventh or twelfth rib | 19 (44) | 61.4 (12.4) | 54.5 (12.8) | 48.1 (12.1) | 42.1 (12.0) | ||
No data 1 | |||||||
Tumor size | |||||||
<7 cm | 21 (62) | 75.9 (9.4) | 70.8 (10.1) | .278 | 66.7 (10.3) | 55.7 (11.2) | .128 |
≥7 cm | 13 (38) | 60.5 (15.4) | 40.4 (15.5) | 41.0 (14.9) | 30.8 (14.3) | ||
No data 10 | |||||||
Primary metastases | |||||||
Absent | 38 (86) | 71.6 (8.0) | 56.2 (9.3) | .173 | 57.6 (8.4) | 43.1 (8.9) | .301 |
Present | 6 (14) | 33.3 (19.2) | 33.3 (19.2) | 33.3 (19.2) | 33.3 (19.2) | ||
Biopsy time point | |||||||
Before 2000 | 19 (43) | 52.6 (11.5) | 47.4 (11.5) | .297 | 47.4 (11.5) | 31.6 (10.7) | .191 |
Afterward | 25 (57) | 79.9 (9.1) | 57.2 (12.9) | 60.0 (10.5) | 52.5 (11.6) | ||
Biopsy with suspicion of malignancy | |||||||
Yes | 32 (76) | 67.1 (9.0) | 53.3 (10.1) | .680 | 57.1 (9.1) | 48.7 (9.6) | .520 |
No | 10 (24) | 67.5 (15.5) | 56.3 (16.5) | 50.0 (15.8) | 30.0 (14.5) | ||
No data 2 | |||||||
Histologic subtype | |||||||
Osteoblastic | 15 (45) | 50.0 (14.4) | 18.8 (15.3) | .056 | 40.7 (13.7) | 15.3 (12.7) | .076 |
Other | 11 (33) | 77.8 (13.9) | 62.2 (17.8) | 72.7 (13.4) | 62.3 (15.0) | ||
Secondary | 7 (21) | 71.4 (17.1) | 71.4 (17.1) | 57.1 (18.7) | 57.1 (18.7) | ||
No data 11 | |||||||
Chemotherapy type | |||||||
Preoperative included | 15 (38) | 50.3 (13.4) | 43.1 (13.2) | .166 | 43.6 (13.3) | 36.4 (12.9) | .221 |
Postoperative only | 25 (43) | 81.3 (8.4) | 63.2 (11.5) | 69.4 (9.7) | 52.3 (11.4) | ||
No surgery 4 | |||||||
Chemotherapy duration | |||||||
≤240 days | 21 (51) | 61.2 (11.6) | 47.2 (12.6) | .753 | 49.5 (11.4) | 42.4 (11.8) | .945 |
>240 days | 20 (49) | 64.7 (11.6) | 51.0 (12.6) | 51.8 (11.7) | 32.9 (11.5) | ||
No data 3 | |||||||
Radiotherapy | |||||||
None documented | 35 (83) | 61.1 (8.8) | 49.3 (9.4) | .188 | 51.7 (8.8) | 36.6 (8.9) | .298 |
Administered | 7 (17%) | 100.0 | 75.0 (21.7) | 71.4 (17.1) | 71.4 (17.1) | ||
No data 2 | |||||||
Surgery of all sites | |||||||
Macroscopically complete | 35 (80) | 77.2 (7.6) | 65.6 (9.0) | <.001 | 68.3 (8.3) | 53.0 (9.4) | - |
Macroscopically incomplete or none | 9 (20) | 14.8 (13.6) | Not reached | Not applicable by definition | |||
Response | |||||||
Good | 5 (33) | 75.0 (21.7) | 75.0 (21.7) | .145 | 75.0 (21.7) | 75.0 (21.7) | .077 |
Poor | 10 (67) | 40.0 (15.5) | 30.0 (14.5) | 30.0 (14.5) | 20.0 (12.6) | ||
Unknown or primary surgery | 29 |
- Note: A total of 44 osteosarcomas of the rib, and the outcome according to potential prognostic variables. Survival estimates are presented together with standard errors. p values of <.05 were considered statistically significant, and those below .1 demonstrated a statistical trend.
DISCUSSION
This report represents the largest analysis of costal osteosarcoma ever presented. It confirms the exceedingly rare involvement of the ribs as primary site of the disease. It demonstrates that a high index of suspicion is required given the overall rarity of the situation. The same biological behavior as well as the same therapeutic and prognostic variables for the much more frequent extremity locations was found to be valid.
Previous reports giving at least some information about this rare site of disease are usually limited to case reports and very few dedicated series, often with a different focus and including, at the very most, a little over 10 patients.3, 4 As reported previously,12, 13 we found only a little more than one in a 100 osteosarcomas originates in the ribs. Our COSS registry mostly caters to patients with extremity primaries.5 Therefore, registration may have been less than complete for those affected at such an unusual tumor location as the rib. Also, not all variables that might be of interest for this particular site may have been collected systematically. We have, however, amassed a large cohort of affected patients and also believe we have captured the most relevant information.
A comparatively high rate of secondary malignancies will have contributed to our patients being several years older than those affected at the more common extremity sites.1 Their median age of 23 years was higher than that we observed in extremity tumors.2 With our group primarily catering to pediatricians5 and therefore more likely to miss older than younger patients, the true median age at which costal osteosarcomas arise is likely to be even higher. The patients’ gender distribution, on the other hand, reflected the situation of osteosarcoma in general, with males more likely to be affected than females.1 The different ribs seemed to be involved at equal proportions. We could not discern any site of predilection. The rate of detectable primary metastatic spread also seemed to broadly reflect the situation of osteosarcoma in general.14
The duration of symptoms before the osteosarcoma diagnosis was similar to that we observed for extremity tumors.2 Both sites have limited amounts of soft tissue hiding neoplastic growth, so lesions are quite easily visible. Nevertheless, more than one-third of the tumors in this series grew to a diameter of more than 7 cm before being noted as such. A high index of suspicion is obviously necessary even at this uncommon site.
It is of concern that the diagnostic surgical procedure was performed without any suspicion of malignancy in almost 25% of our patients and was often a primary resection. As mentioned above, osteosarcomas represented secondary malignancies in approximately one of six of our patients: these secondary tumors were mostly radiation-associated. Patients receiving radiotherapy for cancers of the trunk obviously form a particular risk group in which signs and symptoms must be taken seriously.
Primary costal osteosarcomas in our series were mostly osteoblastic but other subtypes contributed. The formation of neoplastic osteoid should be detectable on a plain x-ray of any osteoblastic lesion and should be searched for diligently.
The thorax cannot be amputated. Therefore, all operated patients received some form of tumor resection. The wide range of surgeons and their frequent unawareness of the margin classification developed for musculoskeletal tumors8 made it impossible to assess these in necessary detail. We therefore had to resort to whether a macroscopically complete resection of all disease sites had been performed. This was the case for just above 75% of patients, considerably less than in the limbs.5 Histologic response information was only available for a small minority of tumors. It was poor in two-thirds of these cases. If confirmed, this would rather reflect the situation of pelvic osteosarcoma, another central tumor,15 than that of extremity lesions.1, 2, 5 Radiotherapy was only administered to seven of our patients, too few to draw any definitive conclusions.
Because our group was primarily centered on chemotherapy protocols,5-7 all patients in this cohort indeed went on to receive some form of systemic treatment. Its median duration was almost 8 months, typical for most recent osteosarcoma regimens. The composition also seemed appropriate for the disease, with most patients receiving at least three of the agents considered most efficacious against the disease.16 Considerably less than half of all patients, however, were treated with any form of preoperative systemic therapy. The choice to use such upfront systemic treatment was not found to be related to knowing the correct diagnosis. This highlights the frequent unfamiliarity of the primarily treating physicians with the potential diagnosis of osteosarcoma and with this tumor's current treatment standards.17 Of the few costal osteosarcomas evaluable, only one-third responded well (<90% viable tumor) to preoperative chemotherapy. This reflects the situation in trunk osteosarcomas in general18-21 but does not at all argue against the use of systemic therapies.
We found that approximately 50% of our patients became 5-year survivors and over 40% became and remained disease-free during this period. This is well below what can be achieved in limb osteosarcoma.1 However, it is somewhat better than what we have previously reported for other central sites.15, 20 The relatively poor outcome compared to limb primaries seems to be related to the presence of some unfavorable prognostic variables, such as an increased proportion of secondary tumors, difficulties in achieving wide surgical margins, and a higher likelihood of receiving inadequate surgery before the correct diagnosis. The abundance of such negative prognostic factors also makes it difficult to assess factors with possible positive implications, such as a good tumor response to upfront chemotherapy. Although this is clearly the largest series of affected patients ever reported, the search for potential prognostic factors is also still impeded by its limited sample size. As a result, relevant questions must remain unanswered. Addressing those would require wide-spread international collaboration. The results, however, give no reason to believe that prognostic factors might differ from those observed in other, more frequent disease sites. We have no clear-cut explanation why osteoblastic lesions may lead to worse outcomes than others.
In summary, our results prove that complete surgery is a mainstay of osteosarcoma therapy even at such an unusual site of disease as the ribs. Affected patients may become long-term, disease-free survivors with added effective multi-agent chemotherapy. Physicians caring for patients with costal lesions should be aware of the differential diagnosis of osteosarcoma to not impede the chance for cure by inappropriate diagnostic or therapeutic measures.
AUTHOR CONTRIBUTIONS
Nicole Bodmer: Conceptualization, data curation, formal analysis, investigation, methodology, validation, and writing. Stefanie Hecker-Nolting: Data curation, investigation, and writing. Godehard Friedel: Data curation and writing. Claudia Blattmann: Funding acquisition, resources, supervision, and writing. Leo Kager: Data curation and writing. Torsten Kessler: Data curation and writing. Matthias Kevric Dok Ass: Data curation, software, and writing. Thomas Kühne: Data curation and writing. Vanessa Mettmann: Data curation and writing. Peter Müller-Abt: Data curation and writing. Benjamin Sorg: Data curation, software, and writing. Matthias Theobald: Data curation and writing. Stefan S. Bielack: Conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, validation, visualization, and writing.
ACKNOWLEDGMENTS
We thank all patients and their legal guardians as well as the dedicated staff of all participating institutions. This study was supported by Förderkreis krebskranke Kinder e. V. Stuttgart.
CONFLICT OF INTEREST STATEMENT
Stefan S. Bielack reports personal fees from Bayer HealthCare, Boehringer Ingelheim, Hoffmann La Roche, MAP Biopharma, and Y-Mabs. Stefanie Hecker-Nolting reports consulting fees from Universitätsspital Basel and grant funding from Eisai and Förderkreis Krebskranke Kinder Stuttgart. Nicole Bodmer reports consulting fees from the University Hospital Zurich. Torsten Kessler reports consulting fees from WWU Muenster. The other authors declare no conflicts of interest.