Ewing Sarcoma in the Pediatric Population: Predictors of Survival Within the United States

Five-Year Survival

Overall, an aggregated 5-year survival rate of 73.92% was found in the entire cohort. Patients with localized cancer had a comparatively improved 5-year survival rate of 84.70% as opposed to those with macrometastatic disease on presentation, with a survival rate of 50.38% (Figure 1).

Open multimedia modal

Patient-Specific Prognostic Factors

Most patients were of adolescent age (57.6%) with the majority being non-Hispanic, White males (Table 1). Most patients had private insurance or other form of managed care (69.23%) followed by government-based insurance (28.19%; Table 1). When separated into alive vs deceased groups, adolescent age at diagnosis was proportionally more common in the deceased group (67.95% vs. 53.86), whereas no notable differences were observed between sex, race, Hispanic status, or insurance type observed (Table 2). Unadjusted patient demographic-specific variables associated with worse survival rates included increasing age (12 to 17 years, HR [95% CI] = 2.27 [1.54 to 3.37]) and Hispanic ethnicity (HR [95% CI] = 1.33 [1.05 to 1.68]), whereas sex, overall race, and insurance status had no notable effect (Table 3). After multivariable analysis, only adolescent age (12 to 17, HR [95% CI] = 2.11 [1.42 to 3.13]) was considered a notable predictor of worse 5-year survival (Figure 2).

Open multimedia modal

Tumor-Specific Prognostic Factors

The most common primary tumor locations were reported within the lower extremities (32.62%) or pelvis (25.57%; Table 1). Approximately 12% of patients were found to have clinically identified macrometastatic disease on presentation. Of the limited cases with the reported largest tumor dimension, most were tumor sizes of <8 cm (AJCC T1 classification). When separated into alive vs. deceased groups, tumor-specific characteristics more commonly reported in the deceased group included pelvic tumor location (37.37% vs. 21.35), larger tumor size (>8 cm: 11.76% vs. 8.89%), and metastasis on presentation (20.69% vs. 10.21%; Table 2). Unadjusted tumor-specific variables associated with worse survival rates included primary tumor location (pelvis, HR [95% CI] = 1.96 [1.57 to 2.46]), larger tumor size (>8 cm, HR [95% CI] = 2.51 [1.68 to 3.75]), and presence of metastasis at the time of diagnosis (HR [95% CI] = 2.33 [1.87 to 2.91]; Table 3), whereas grade had no notable effect. Primary tumor locations involving facial and skull or mandible-specific locations had lower mortality rates. After multivariable analysis, larger tumor size (>8 cm, HR [95% CI] = 2.24 [1.49 to 3.35]) and positive metastases on presentation (HR [95% CI] = 2.15 [1.72 to 2.70]) were associated with considerable significance in predicting worse 5-year survival (Figure 2).

Treatment-Specific Prognostic Factors

Most patients underwent treatment with a sequenced chemotherapy and surgical resection; approximately 20% of patients had received some degree of radiation treatment (Figure 1). When separated into alive vs deceased groups, treatment-specific characteristics more commonly reported in the deceased group included no primary surgery being performed, whereas no differences were observed in chemotherapy alone or radiation alone treatments (Table 2). Alive patients were reported to undergo combined chemotherapy and surgery at a markedly higher rate than deceased patients (60.76% versus 39.35%, respectively), who more often underwent nonsurgical management. Unadjusted treatment-specific variables associated with worse survival rates were positive margins after surgical resection (HR [95% CI] = 1.71 [1.19 to 2.47]), tumors in which no primary site surgery was performed (HR [95% CI] = 2.66 [2.13 to 3.30]), and patients treated without chemotherapy or surgery (HR [95% CI] = 3.28 [2.13 to 5.05]; Table 3). After multivariable analysis, only positive margin status (HR [95% CI] = 1.70 [1.18 to 2.46]) was considered significant in predicting worse 5-year survival (Figure 2).

Survival Rates

Overall, an aggregated 5-year survival rate of 74.5% was found among all included patients. Patients with localized disease had a comparatively improved 5-year survival rate of 84.70% as opposed to patients with macrometastatic disease on presentation with a survival rate of 50.4% (Figure 1). These findings are consistent, albeit slightly higher, compared with similar epidemiology-based and review studies for localized (approximately 70% to 80%) and metastatic (approximately 30% to 40%) disease.^17,20,29-33 However, many of these larger population-driven and hospital database–driven reports are 1 to 2 decades old. Therefore, the authors believe that the improved survival rates reported in this study are likely a result of gradual improvement/emphasis on multidisciplinary care provided at larger tertiary referral centers with greater access to randomized control trials.^34-37

Patient-Specific Prognostic Factors

Increasing age was associated with overall worse survival rates, with adolescents demonstrating worse 5-year survival rates compared with younger age groups. This is consistent with previous findings that have shown the survival rate of ES to be inversely proportional to age among pediatric and young adult populations.^38-40 Ewing tumor size and location have some purported age dependence with adolescents having markedly larger tumors and higher incidence of occurrence within the pelvis and axial skeleton as compared with preadolescent patients.⁴¹ Given the worse prognosis associated with axial tumors,⁴² this was likely a contributor to worse outcomes with increasing age in this analysis.

Diagnosis of ES primarily occurred in non-Hispanic White males, which is consistent with ES's historical predilection for European ancestries and strikingly low incidence in populations of African descent.¹⁵ This is likely secondary to genetic germline variants that are protective for African Americans.⁴³ However, there is a tumor dedifferentiation risk of ES in people of African descent, which may be a result of a specific genetic alteration within the EGR2 gene.⁴⁴ Regardless, patient sex and race did not account for notable differences in mortality status.

Tumor-Specific Prognostic Factors

As expected, larger tumor size and metastatic disease on presentation were both associated with worse overall survival rates. These findings are consistent with tumor staging through the AJCC updated staging guidelines for primary bone tumors.⁴⁵ Importantly, the most updated AJCC staging system (eighth edition) separates pelvis and spinal primary bone tumors to better reflect the overall difference in survival or treatments, respectively, as compared with appendicular tumors.^42,45 It should also be noted that this study's analysis showed notable differences in tumor location among deceased and nondeceased patients diagnosed with ES, with deceased patients having notably higher percentages of pelvis locations (Tables 2 and 3). Axial primary bone malignancies, particularly within the pelvis, have been convincingly proven to have worse long-term survival rates compared with appendicular tumors,^46-53 hence the updates to the AJCC staging system. It is uncertain as to why pelvic tumor location did not remain a notable predictor of worse mortality after the adjusted multivariable analysis (Table 3 versus Figure 2); however, this may reflect improvements in targeted radiation and chemotherapy treatments. Overall, improvements in targeted adjuvant treatments and the emphasis on multidisciplinary care are likely the two drivers behind the improved metastatic disease and/or pelvic tumor location survival rates.

The updated AJCC also lumps the appendicular skeleton with trunk, skull, and facial bone tumors for classification purposes of primary bone tumors. This study found that facial and mandible-specific primary Ewing tumor locations have markedly improved survival compared with appendicular tumors. These facial and mandible-specific primary tumors may be more easily noticed by patients and parents and therefore diagnosed earlier and treated in a timelier manner. However, these findings appear to contrast with similar survival rates to contemporary appendicular tumors.^54-56 Although unclear for the causality, facial and mandible-specific tumors may need to be considered for separate categorization in newer editions of the AJCC Cancer Staging Manual.

Treatment-Specific Prognostic Factors

Before the 1970s, amputation was the only believed methodology for managing local tumor burden.⁵⁷ The shift to combined chemotherapy with limb-preservation surgery revolutionized outcomes and patient functionality.^58,59 As expected, mortality was found to be markedly affected by treatment type, with chemotherapy and surgery combination treatment associated with improved 5-year survival rates in this analysis. Of note, limb-preservation surgery is contingent on the tumor's candidacy for resection and the patient/parents' desire for that surgery. For example, large and invasive pelvic tumors or metastatic burdens may preclude patients from undergoing combined chemotherapy and surgical management. These patients may be managed with isolated radiation treatments or combined with chemotherapy for palliative treatment. In this investigation, ES patients treated with likely isolated palliative radiation showed a trend toward improved survival compared with no treatment. We believe that this was likely unresectable tumors in central locations such as the previously mentioned pelvic location subgroup.

Among patients who undergo tumor resection, survival rates have been predicated on margin status and tumor response to chemotherapy. Similarly, this study showed that positive margins after surgical resection were associated with lower 5-year survival rates. Unfortunately, there is considerable variability in the methodology of reporting margins.⁶⁰ For example, standard teaching stressed the importance of a 3-cm bony margin and soft-tissue layer surrounding the tumor.⁶¹ However, less stringent 1.5-cm margins have demonstrated equivalent oncologic outcomes in osteosarcoma.⁶² Despite these controversies, the basic principle of reducing tumor burden and local recurrence through negative margins upon wide resection persists.^19,50,63,64 Histologic response to chemotherapy is important in ES⁴⁶; however, this variable is not reported in the NCDB. Providers should bear in mind that even with successful treatment and remission past the 5-year mark, patients may require continued monitoring from their regular oncologist with referral to the musculoskeletal oncology surgeon as needed. Even with curative treatment without recurrence, there are notable short- and long-term treatment-related complications as a result of the chemotherapy and resection of these malignant bone tumors.^65,66

Limitations

There are several limitations to the use of large databases. Although the NCDB captures most new cancer patients in the United States^27,28 and displays a high rate of 5-year follow-up,²⁶ it is ultimately a hospital-based data set and not population based (such as Surveillance, Epidemiology, and End Results data). Therefore, patient inclusion is specific only to patients treated by hospitals accredited by the American College of Surgeons Committee on Cancer. Furthermore, analysis is limited to variables provided, which often results in the lack of specific data granularity. This includes exact tumor location, lack of EWSR1 status for diagnostic confirmation, histologic response to chemotherapy, radiation dosing, type of chemotherapy received, or patients' involvement in clinical trials. The NCDB seeks to be a comprehensive database consistent in reporting variables among tumors. However, this may lead to errors in reporting or erroneous variables such as reporting various grades of ES of bone, which is generally universally considered undifferentiated.

In addition, the NCDB is unable to link separate or associated cancer diagnoses for the same patient, which may serve pertinent in patients with genetic predispositions. It reports each individual cancer diagnosis as a different case. However, researchers can exclude patients indicated to have a history of other cancers, thereby minimizing the effect of prior malignant neoplasms, as was done in this analysis. Furthermore, we were unable to report on survival rates regarding patients who initially presented with localized disease with later progression to metastatic disease despite having received treatment. Finally, this analysis was only able to report overall survival rather than event-free survival rates as recurrence and specific complications are not explicitly provided by the NCDB.

Looking Forward—Treatments on the Horizon

Compressed chemotherapy induction with vincristine, doxorubicin, and cyclophosphamide remains the standard for ES within the United States.^67,68 Although various trials continue, major changes in chemotherapy toward either malignancy have not been implemented in the United States in a decade. Current efforts in novel treatments include investigations into stereotactic radiosurgery, proton beam therapy, and even some targeted immunotherapies.⁶⁹ Low incidence rates and difficulty obtaining adequately powered clinical trials remain burdensome for these investigations.