Impact of radiotherapy on survival outcomes in metastatic neuroblastoma a propensity score matched SEER database analysis

Abstract

Metastatic neuroblastoma is a highly aggressive pediatric malignancy with poor prognosis. Radiotherapy is commonly used as part of multimodal treatment, but its impact on survival outcomes remains controversial. This study investigates the association between radiotherapy and survival in patients with metastatic neuroblastoma using data from the Surveillance, Epidemiology, and End Results (SEER) database. A retrospective analysis was conducted on 4,850 patients diagnosed with metastatic neuroblastoma from the SEER database (2004–2015). After applying exclusion criteria, 981 patients were included, with 368 receiving radiotherapy. Propensity score matching (PSM) was used to balance baseline characteristics between the radiotherapy and non-radiotherapy groups, resulting in 234 patients in each group. Survival outcomes were analyzed using Kaplan–Meier curves and Cox proportional hazards models. Before PSM, no significant difference in overall survival (OS) and cancer-specific survival (CSS) was observed between patients who received radiotherapy and those who did not. After PSM, there was a trend toward improved OS and CSS in the radiotherapy group, though statistical significance was not reached. Cox regression analysis identified age ≥ 1 year and non-adrenal primary tumor site as significant independent predictors of poorer OS and CSS. Radiotherapy was not an independent predictor of survival in the multivariate analysis, but a trend toward a survival benefit was noted, particularly in patients with larger tumors and those who underwent surgery. Radiotherapy may improve survival in metastatic neuroblastoma patients with large tumors or surgical resection, though its independent benefit remains uncertain. Personalized strategies require integrating updated COG risk stratification with biomarkers and prospective trials assessing long-term outcomes to refine treatment approaches.

Introduction

Neuroblastoma, an embryonal tumor originating primarily from the autonomic nervous system, includes neuroblastoma, ganglioneuroblastoma, and ganglioneuroma. Notably, neuroblastoma comprises 97% of these tumors, characterized by heterogeneity in site of occurrence, histopathological features, and biological properties¹. It ranks among the most common extracranial solid tumors in children and is the most prevalent cancer type in infants under 12 months^2,3. Predominantly affecting children, neuroblastoma accounts for approximately 15% of all childhood cancer deaths, with an incidence rate of 107 cases per million⁴. The onset is closely associated with age, with a median diagnosis age of 17.3 months, and 40% of patients are diagnosed before the age of one. The 5-year survival rate is estimated to range from 75 to 98% for children with low- and intermediate-risk neuroblastoma, while it falls below 50% for those with high-risk neuroblastoma⁵. Recurrence or progression at metastatic sites remains the leading cause of neuroblastoma-related mortality⁶.

Radiation therapy is a standard treatment for high-risk neuroblastoma, typically administered following consolidation therapy to prevent local tumor recurrence^7,8. However, debate persists regarding whether the radiation field should encompass lymph nodes adjacent to the primary tumor^9,10. Similarly, there is uncertainty about which metastatic lesions should be irradiated and how this affects the risk of local and overall recurrence^11,12. Conflicting data exist on the efficacy of radiation therapy for patients with residual tumors at the primary site post-consolidation therapy. Although radiation therapy has demonstrated some efficacy in treating high-risk neuroblastoma, there is no level one evidence supporting its optimal use^13,14. Additionally, there is a scarcity of clinical studies on metastatic neuroblastoma in children, particularly large-scale studies focused on survival prognosis. This gap in research has impeded the development of mature treatment strategies for metastatic neuroblastoma, highlighting the critical need for more foundational and clinical studies to provide patients with more precise and personalized treatment options.

This study aims to leverage the Surveillance, Epidemiology, and End Results (SEER) database and apply propensity score matching (PSM) analysis to evaluate the overall and specific effects of radiation therapy on the prognosis of metastatic neuroblastoma. The objective is to generate evidence that will inform clinical decision-making and provide valuable reference data for the design of future clinical trials.

Materials and methods

Patients and variables

Specific clinicopathological data and prognostic outcomes of neuroblastoma patients from 2004 to 2015 were retrieved from the SEER database. Ethical approval or local statements were not required for this study, as all data utilized were obtained through publicly accessible methods from the SEER database.

Data were obtained for patients diagnosed with neuroblastoma according to the International Classification of Diseases for Oncology (ICD-O-3) histology codes: 9490 (ganglioneuroblastoma), 9500 (neuroblastoma, not otherwise specified), 9504 (spongioneuroblastoma), and 9522 (olfactory neuroblastoma). The following variables were extracted: age at diagnosis, gender, race, year of diagnosis, primary site, tumor grade (defined as well-differentiated I–II, poorly differentiated III, and undifferentiated IV), tumor size, histology, chemotherapy, surgery, radiation therapy, metastasis, survival time, and cause of death. Exclusion criteria included: (1) age ≥ 15 years; (2) diagnosis not confirmed by positive histology; (3) summary stage categorized as blank, regional, localized, unknown, or unstaged; (4) absence of distant metastasis, or unknown/blank metastasis status. The study flowchart is shown in Fig. 1, and all included patients (n = 919) were divided into two groups based on whether they received radiation therapy.

Fig. 1.

Flow chart for screening patients.

Statistical analysis

Baseline characteristics of patients and treatments were summarized using descriptive statistics. Differences in clinical characteristics between the radiotherapy and non-radiotherapy groups were analyzed with chi-square and Fisher’s exact tests, as appropriate. To minimize selection bias and ensure comparability between the radiation therapy group and the non-radiation therapy group, PSM was applied. The matching variables included age, gender, race, primary tumor site, tumor grade, tumor size, histological type, chemotherapy status, and surgical intervention. A 1:1 nearest neighbor algorithm without replacement was used, resulting in 234 matched pairs of patients.

Kaplan–Meier survival curves were utilized to analyze survival outcomes, including overall survival (OS) and cancer-specific survival (CSS), with differences between groups assessed via the log-rank test. Cox proportional hazards models were applied to identify independent prognostic factors for survival, with results expressed as hazard ratios (HR) and 95% confidence intervals (CI). Both univariate and multivariate analyses were conducted, with variables showing significance (P < 0.05) in univariate analysis included in the multivariate model. Statistical analyses were performed using R software (version 4.3.0), with a P-value of less than 0.05 considered statistically significant.

Results

Baseline patient characteristics

Among the 919 patients diagnosed with metastatic neuroblastoma in the SEER database from 2004 to 2015, 368 received radiation therapy, while 551 did not. The demographic characteristics of patients with or without radiation therapy are summarized in Table 1. Overall, no significant associations were observed with gender, tumor grade, or histological type. However, statistically significant differences were noted for age (P < 0.001), race (P = 0.013), primary site (P = 0.049), tumor size (P < 0.001), chemotherapy (P < 0.001), and surgery (P < 0.001). After PSM, the radiation therapy group and the non-radiation therapy group each comprised 234 matched patients. No significant associations were observed between variables in the two groups after PSM (Table 1).

Table 1.

Characteristics of radiation therapy in patients afflicted with metastatic neuroblastoma.

Variable	Group	Before PSM		P_value	After PSM		P_value
		Radiotherapy N (%)	Non-radiotherapy N (%)		Radiotherapy N (%)	Non-radiotherapy N (%)
Age (years)	< 1	37 (10.1)	225 (40.8)	< 0.001	35 (15)	32 (13.7)	0.792
	≥ 1	331 (89.9)	326 (59.2)		199 (85)	202 (86.3)
Sex	Male	212 (57.6)	305 (55.4)	0.544	129 (55.1)	144 (61.5)	0.189
	Female	156 (42.4)	246 (44.6)		105 (44.9)	90 (38.5)
Race	White	258 (70.1)	417 (75.7)	0.013	171 (73.1)	174 (74.4)	0.601
	Black	73 (19.8)	70 (12.7)		40 (17.1)	33 (14.1)
	Others	37 (10.1)	64 (11.6)		23 (9.8)	27 (11.5)
Primary site	Adrenal	250 (67.9)	338 (61.3)	0.049	159 (67.9)	167 (71.4)	0.482
	Others	118 (32.1)	213 (38.7)		75 (32.1)	67 (28.6)
Grade	Grades I–II	4 (1.1)	6 (1.1)	0.907	4 (1.7)	1 (0.4)	0.298
	Grades III–IV	215 (58.4)	312 (56.6)		135 (57.7)	145 (62)
	Unknown	149 (40.5)	233 (42.3)		95 (40.6)	88 (37.6)
Tumor Size	< 5 cm	61 (16.6)	119 (21.6)	< 0.001	40 (17.1)	39 (16.7)	0.906
	≥ 5 cm	237 (64.4)	273 (49.5)		144 (61.5)	141 (60.3)
	Unknown	70 (19.0)	159 (28.9)		50 (21.4)	54 (23.1)
Histologic type	Ganglioneuroblastoma	26 (7.1)	36 (6.5)	0.899	18 (7.7)	18 (7.7)	1.000
	Neuroblastoma	341 (92.7)	514 (93.3)		216 (92.3)	216 (92.3)
	Olfactory neuroblastoma	1 (0.3)	1 (0.2)		0 (0.0)	0 (0.0)
Chemotherapy	Yes	366 (99.5)	501 (90.9)	< 0.001	232 (99.1)	232 (99.1)	1.000
	No	2 (0.5)	50 (9.1)		2 (0.9)	2 (0.9)
Surgery	Yes	322 (87.5)	327 (59.3)	< 0.001	200 (85.5)	185 (79.1)	0.090
	No	46 (12.5)	224 (40.7)		34 (14.5)	49 (20.9)

Survival analysis for OS and CSS

Before PSM, there were no significant differences in OS and CSS between patients who received radiation therapy and those who did not (P > 0.05), as shown in Fig. 2A and C.

Fig. 2.

Comparison of OS and CSS between the radiation therapy and the non-radiation therapy groups. (A) OS before PSM; (B) OS after PSM. (C) CSS before PSM; (D) CSS after PSM.

After 1:1 matching, analysis of OS and CSS between the 234 patients in the radiation therapy group and the 234 patients in the non-radiation therapy group also showed no significant statistical differences (P > 0.05), as seen in Fig. 2B and D.

However, when stratifying tumor patients by age, primary site, tumor differentiation grade, tumor size, chemotherapy, and surgery, it was found that OS was significantly better in the radiation therapy group compared to the non-radiation therapy group for patients aged ≥ 1 year (P = 0.009, Fig. 3B), patients with tumor differentiation Grades I–II (P = 0.046, Fig. 3E), patients with tumor size ≥ 5 cm (P = 0.001, Fig. 4B), patients who received chemotherapy (P = 0.042, Fig. 4C), and patients who underwent surgery (P = 0.040, Fig. 4E). However, there were no differences in OS between the radiation therapy and non-radiation therapy groups for patients aged < 1 year, with tumors in the primary site, with Grades III–IV differentiation, with tumor size < 5 cm, without chemotherapy, and without surgery (Figs. 3A, C, D, F, 4A, D, and F).

Fig. 3.

Comparison of OS between radiotherapy and non-radiotherapy groups after stratifying tumor patients by age, primary site, tumor differentiation grade, tumor size, chemotherapy and surgery.

Fig. 4.

Comparison of OS between radiotherapy and non-radiotherapy groups after stratifying tumor patients tumor size, chemotherapy and surgery.

Prognostic factors in metastatic neuroblastoma

Cox regression analysis assessed the impact of various factors on OS and CSS in patients with metastatic neuroblastoma, as detailed in Tables 2 and 3. Multivariate analysis revealed that patients aged 1 year or older had a significantly higher risk of mortality compared to those under 1 year, with HRs for OS at 1.918 (95% CI 1.185–3.104, P = 0.008) and similar results for CSS (HR = 1.918, 95% CI 1.185–3.104, P = 0.008). Additionally, the primary tumor site was identified as a crucial factor influencing survival outcomes. After adjusting for other variables, patients with tumors located outside the adrenal gland demonstrated higher survival rates for both OS (HR = 0.680, 95% CI 0.501–0.922, P = 0.013) and CSS (HR = 0.673, 95% CI 0.493–0.920, P = 0.013).

Table 2.

Univariate and multivariate Cox regression model analysis of OS between the Radiotherapy group and the Non-radiotherapy group.

Variable	Group	Univariate analysis		Multivariate analysis
		Hazard ratio (95% CI)	P_value	Hazard ratio (95% CI)	P_value
Age(years)	< 1	Reference		Reference
	≥ 1	1.833 (1.144–2.937)	0.012	1.918 (1.185–3.104)	0.008
Sex	Female	Reference		Reference
	Male	0.981 (0.748–1.286)	0.888	0.974 (0.744–1.274)	0.845
Race	White	Reference		Reference
	Black	1.074 (0.738–1.565)	0.709	1.122 (0.784–1.606)	0.529
	Others	1.471 (0.977–2.214)	0.065	1.487 (0.984–2.246)	0.060
Primary site	Adrenal	Reference		Reference
	Others	0.753 (0.556–1.019)	0.066	0.680 (0.501–0.922)	0.013
Grade	Grades I–II	Reference		Reference
	Grades III–IV	2.873 (0.402–20.555)	0.293	3.402 (0.468–24.722)	0.226
	Unknown	3.029 (0.422–21.750)	0.271	3.372 (0.465–24.470)	0.229
Tumor size	< 5 cm	Reference		Reference
	≥ 5 cm	1.303 (0.857–1.982)	0.216	1.262 (0.827–1.926)	0.280
	Unknown	1.397 (0.870–2.245)	0.167	1.302 (0.803–2.111)	0.285
Histologic Type	Neuroblastoma	Reference		Reference
	Ganglioneuroblastoma	0.937 (0.563–1.560)	0.802	0.950 (0.565–1.596)	0.846
	Olfactory neuroblastoma	NA	NA	NA	NA
Chemotherapy	Yes	Reference		Reference
	No	0.502 (0.070–3.580)	0.492	0.597 (0.081–4.377)	0.612
Surgery	Yes	Reference		Reference
	No	1.361 (0.968–1.913)	0.076	1.407 (0.969–2.042)	0.073
Radiotherapy	Yes	Reference		Reference
	No	1.295 (0.990–1.694)	0.059	1.275 (0.974–1.669)	0.077

Significant values are in bold.

Table 3.

Univariate and multivariate Cox regression model analysis of CSS between the Radiotherapy group and the Non-radiotherapy group.

Variable	Group	Univariate analysis		Multivariate analysis
		Hazard ratio (95% CI)	P_value	Hazard ratio (95% CI)	P_value
Age (years)	< 1	Reference		Reference
	≥ 1	1.833 (1.144–2.937)	0.012	1.918 (1.185–3.104)	0.008
Sex	Female	Reference		Reference
	Male	0.981 (0.748–1.286)	0.888	0.954 (0.726–1.256)	0.739
Race	White	Reference		Reference
	Black	1.074 (0.738–1.565)	0.709	1.090 (0.744–1.597)	0.659
	Others	1.471 (0.977–2.214)	0.065	1.469 (0.965–2.237)	0.073
Primary site	Adrenal	Reference		Reference
	Others	0.753 (0.556–1.019)	0.066	0.673 (0.493–0.920)	0.013
Grade	Grades I–II	Reference		Reference
	Grades III–IV	2.873 (0.402–20.555)	0.293	3.327 (0.458–24.185)	0.235
	Unknown	3.029 (0.422–21.750)	0.271	3.347 (0.461–24.302)	0.232
Tumor size	< 5 cm	Reference		Reference
	≥ 5 cm	1.303 (0.857–1.982)	0.216	1.237 (0.809–1.891)	0.327
	Unknown	1.397 (0.870–2.245)	0.167	1.261 (0.775–2.052)	0.350
Histologic type	Neuroblastoma	Reference		Reference
	Ganglioneuroblastoma	0.937 (0.563–1.560)	0.802	0.982 (0.583–1.653)	0.945
	Olfactory neuroblastoma	NA (NA–NA)	NA	NA (NA–NA)	NA
Chemotherapy	Yes	Reference		Reference
	No	0.502 (0.070–3.580)	0.492	0.589 (0.080–4.320)	0.603
Surgery	Yes	Reference		Reference
	No	1.361 (0.968–1.913)	0.076	1.462 (1.001–2.137)	0.049
Radiotherapy	Yes	Reference		Reference
	No	1.295 (0.990–1.694)	0.059	1.278 (0.970–1.682)	0.081

Significant values are in bold.

In the multivariate model for OS (HR = 1.275, 95% CI 0.974–1.669, P = 0.077) or CSS (HR = 1.278, 95% CI 0.970–1.682, P = 0.081), radiation therapy itself did not reach statistical significance as an independent prognostic factor. However, the analysis suggests a trend toward survival benefit from radiation therapy, especially when considering the marginal P-values. These results highlight the complexity of treatment decisions for metastatic neuroblastoma, where multiple factors, including age and tumor location, play a crucial role in patient prognosis.

Discussion

This study provides critical insights into the role of radiotherapy in treating metastatic neuroblastoma, emphasizing its potential impact on survival outcomes. The major findings suggest a trend toward improved OS and CSS in patients who received radiotherapy after propensity score matching, though these results did not achieve statistical significance. Subgroup analyses indicated that radiotherapy may confer a survival benefit, particularly in patients with larger tumors and those who underwent surgery. However, in the multivariate Cox regression analysis, radiotherapy did not emerge as an independent predictor of survival, implying that its benefits may be context-dependent, influenced by factors such as tumor size and surgical intervention. These findings highlight the complexity of treating metastatic neuroblastoma and the necessity for individualized treatment strategies that account for the unique characteristics of each patient’s disease.

The findings of our study align with and build upon previous research investigating the impact of radiotherapy on survival outcomes in neuroblastoma. Consistent with our results, several studies have indicated that radiotherapy may confer a survival advantage in specific subgroups of neuroblastoma patients, particularly those with residual disease post-surgery or larger primary tumors^15–17. For example, Tesson¹⁸ underscored the potential benefits of radiotherapy in improving local control, thereby potentially enhancing overall survival, especially in high-risk cases. Our study supports these findings, demonstrating a trend toward improved survival in patients receiving radiotherapy, although a statistically significant difference was not observed in the overall cohort after PSM.

However, our study differs from some earlier reports in its emphasis on the lack of independent prognostic significance of radiotherapy in multivariate analysis. While previous studies, such as those by De et al.¹⁹, have highlighted the potential of radiotherapy to significantly improve outcomes in certain patient populations, our findings suggest that its benefits may be more context-specific, influenced by factors like tumor size and surgical intervention. Moreover, unlike other studies that have demonstrated clear survival benefits with radiotherapy across broader patient populations^20,21, our results indicate that these benefits may not be universally applicable, underscoring the need for personalized treatment approaches in metastatic neuroblastoma.

A key strength of this study is the use of PSM to minimize selection bias, providing a more accurate assessment of radiotherapy’s impact on survival outcomes in metastatic neuroblastoma^22,23. PSM facilitated the creation of well-balanced groups with comparable baseline characteristics, addressing a significant limitation in many previous studies that lacked such rigorous matching procedures²⁴. This methodological approach allowed us to more effectively isolate the effect of radiotherapy, reducing the influence of confounding variables such as age, tumor size, and extent of surgical intervention. Despite this robust design, our study found that radiotherapy was not an independent predictor of survival in the multivariate analysis, challenging the conventional wisdom that suggests a universal benefit of radiotherapy in this patient population²⁵. These findings imply that while PSM is a powerful tool for observational studies, the complex nature of metastatic neuroblastoma treatment outcomes may require even more nuanced approaches, potentially including stratified or subgroup analyses, to fully understand the contexts in which radiotherapy is most beneficial²⁶.

Despite the strengths of our study, several limitations must be acknowledged. First, as a retrospective analysis using data from the SEER database, the study is inherently subject to the biases and limitations associated with observational research, including potential inaccuracies in data reporting and coding²⁷. Additionally, the SEER database lacks detailed information on treatment protocols, such as radiation dose and field, which are critical for accurately assessing the true impact of radiotherapy on survival outcomes¹. Furthermore, the study population was limited to patients under 15 years of age, which may restrict the generalizability of the findings to older populations typically affected by neuroblastoma²⁸. This study was based on data from 2004 to 2015 and utilized the older COG risk classification system, which categorized all patients with International Neuroblastoma Staging System (INSS) stage 4 as high-risk. However, the updated COG risk classification system²⁹ has significantly refined risk stratification by integrating the International Neuroblastoma Risk Group Staging System (INRGSS) and chromosomal abnormalities, such as 1p/11q deletions. For instance, metastatic patients without MYCN amplification or segmental chromosomal aberrations (SCA) may now be reclassified as intermediate- or low-risk, potentially reducing the need for radiotherapy. This heterogeneity in risk grouping may partially explain the lack of overall survival benefit observed in our study. Future studies should incorporate molecular biomarkers to prospectively validate the role of radiotherapy in patients with true high-risk features such as MYCN amplification or SCA.

Future studies should aim to overcome these limitations by incorporating prospective designs with more detailed clinical data, including specifics of radiotherapy treatment and its timing relative to other modalities. Additionally, research exploring the molecular and genetic factors influencing the response to radiotherapy in metastatic neuroblastoma is needed³⁰. Such studies could pave the way for personalized treatment approaches, optimizing the use of radiotherapy in conjunction with other therapeutic strategies³¹. Finally, examining the impact of radiotherapy on quality of life and long-term survivorship in this patient population would provide a more comprehensive understanding of its role in neuroblastoma treatment³².

Conclusion

This study highlights the nuanced role of radiotherapy in the treatment of metastatic neuroblastoma. While radiotherapy did not demonstrate an independent survival benefit in the overall cohort, subgroup analyses suggested potential clinical utility in patients with larger tumor burdens or those undergoing surgical intervention. The limitations of the older risk stratification system underscore the need for more precise definitions of radiotherapy indications. Future research should integrate the revised COG risk classification system with molecular biomarkers to better identify patient subgroups that may derive the greatest benefit from radiotherapy. Additionally, prospective studies incorporating detailed clinical and molecular data are essential to refine therapeutic strategies and improve outcomes for patients with this aggressive malignancy. Further exploration of the long-term effects of radiotherapy on survival and quality of life will also be critical to informing patient-centered treatment approaches.

Author contributions

SXL proposed the conception, designed the study and supervised the whole analysis. AA acquired the data and performed statistical analysis. FA drafted the article. KL contributed to the data acquisition and checked the statistical analysis.

Funding

Supported by Natural Science Foundation of Xinjiang Uygur Autonomous Region (Grant number: 2024D01A106).

Data availability

Publicly available datasets were analyzed in this study. The data can be found here: https://seer.cancer.gov/data/

Declarations

Competing interests

The authors declare no competing interests.