Association of image-defined risk factors with clinical features in thoracic neuroblastoma and the development of a prognostic prediction model.

Introduction

Neuroblastoma is one of the most common extracranial malignant tumors in pediatric populations, predominantly affecting children under 5 years old. It accounted for approximately 8%–10% of all childhood malignancies ( 1 ). The disease arises from primordial neural crest cells, with the most frequently affected anatomical sites being the adrenal medulla, retroperitoneum, mediastinum, and cervical region ( 2 , 3 ). Neuroblastoma exhibits significant heterogeneity in clinical outcomes. While a minority of cases may experience spontaneous tumor regression, the majority of pediatric patients faced a poor prognosis. Consequently, neuroblastoma contributed to approximately 12% of cancer-related deaths among children under 15 years old ( 4 – 6 ).

The aggressive biological behavior of neuroblastoma can increase the surgical risk and therapeutic challenges. To address this, the International Neuroblastoma Risk Group (INRG) introduced the International Neuroblastoma Risk Group Staging System (INRGSS) in 2009, which was based on image-defined risk factors (IDRFs) ( 7 ). This staging system did not depend on surgical findings and enabled preliminary staging of patients before pre-treatment, thereby facilitating early implementation of individualized therapeutic strategies.

However, the INRGSS classification system was limited to classifying tumors into stages L1 and L2 based on the presence or absence of IDRFs. There remained a lack of consensus regarding the prognostic significance of different types and quantities of IDRFs. Given the considerable variation in the incidence of neuroblastoma across anatomical locations, most clinical investigations had concentrated on adrenal and abdominal tumors, with limited systematic studies focusing on thoracic neuroblastoma. Existing evidence suggested that thoracic neuroblastoma exhibited distinct clinical and tumor biological characteristics compared to abdominal neuroblastoma, as well as differences in the characteristics and incidence of IDRFs ( 8 , 9 ). This study aimed to retrospectively analyze clinical data from a single institution to characterize the clinical features, IDRFs distribution, and prognostic outcomes of pediatric patients with thoracic neuroblastoma. Furthermore, we evaluated the association between IDRFs and clinical features, and assessed their potential prognostic value, with the objective of providing clinical evidence to support the validation and potential refinement of the INRGSS staging system.

Materials and methods

Study population

This study retrospectively collected data from patients with thoracic neuroblastoma treated at the Children's Hospital of Nanjing Medical University between January 2017 and June 2024. Before the operation, a comprehensive evaluation of the tumor's relationship with surrounding tissues and organs should be conducted based on contrast-enhanced CT and/or MRI, as well as other imaging modalities. For those who could not undergo single-stage tumor resection, neoadjuvant chemotherapy was given, and the postponed or second-stage surgery was determined based on the extent of tumor regression. The inclusion criteria for this study were as follows: (1) age < 18 years, with no sex restrictions; (2) the primary tumor was located in the thoracic region; (3) postoperative pathology confirmed the diagnosis of neuroblastoma; (4) complete clinical information, imaging data, and follow-up records. This study has been approved by the Hospital Ethics Committee of the Children's Hospital of Nanjing Medical University (Permission No. 202205047-1).

Data collection

The sample size for this study was determined based on the availability of data within our dataset. The collected clinical variables included the sex, age at initial diagnosis, and the first available preoperative serum levels of neuron-specific enolase (NSE), lactate dehydrogenase (LDH), total protein (TP), high-density lipoprotein (HDL), and hemoglobin concentration (Hb). Imaging data encompassed preoperative and postoperative CT and MRI scans. These imaging data were independently reviewed by two trained professionals to assess IDRFs. Discrepancies or indeterminate imaging features were resolved through consensus evaluation by experienced senior radiologists. The maximum tumor diameter (MTD), defined as the longest dimension of the tumor in its largest cross-sectional image, was recorded as the primary tumor size. Pathological classification was performed in accordance with the International Neuroblastoma Pathology Classification (INPC) system ( 12 ), based on postoperative histopathological findings. Tumor staging was conducted using the International Neuroblastoma Staging System (INSS) to determine the postoperative clinical stage.

In 2009, the International Neuroblastoma Risk Group (INRG) described more than twenty types of IDRFs occurring in various anatomical regions of the human body. Among these, Delforge et al. summarized the IDRF of thoracic neuroblastoma and classified them into four major categories: infiltrative, vascular, neurological, and extensive ( 13 ).

Overall survival (OS) was defined as the interval from the date of diagnosis to either the date of death or the last follow-up. Event-free survival (EFS) was defined as the time from diagnosis to the occurrence of the first adverse event, including tumor progression, recurrence, secondary malignancies, or death from any cause, with censored observations determined by the date of the last follow-up. Patients who were lost to follow-up were classified based on their last documented clinical status, with those still alive at the time of the last follow-up considered as censored observations. The follow-up endpoint for this study was set at December 2024.

Statistical analysis

Data were processed, analyzed, and visualized using SPSS (version 26.0) and R software (version 4.3.3). Continuous variables were assessed using independent-sample t -test or Mann–Whitney U test, while categorical variables were evaluated using the Chi-square ( χ 2 ) test or Fisher exact test. p < 0.05 was considered statistically significant. The cut-off values of the laboratory variables and the maximum tumor diameter determined in this study were achieved by using the R package “Cutoff” (version 1.3). Univariate logistic regression analysis was initially conducted to identify potential factors associated with IDRFs. Variables with p < 0.10 were subsequently incorporated into a bidirectional stepwise multivariate logistic regression model to identify independent predictors. Survival and prognostic data were analyzed using the Kaplan–Meier method and Log-rank test. Univariate and multivariate survival analyses were performed using the Cox proportional hazards model. Independent risk factors influencing event-free survival (EFS) were selected based on the minimum Akaike Information Criterion (AIC) and included in the development of nomogram model. Given the extremely low proportion of N-MYC amplification in this study cohort ( n = 2), this well-established prognostic biomarker was excluded from the multivariable prognostic analysis to avoid model instability and overfitting.

Nomogram, also referred to as a nomograph, is a widely used graphical tool that enables the estimation of individual probabilities of disease occurrence or prognosis ( 14 , 15 ). The discriminative ability of the nomogram was evaluated using the concordance index (C-index) and time-dependent area under the receiver operating characteristic curve (AUC). In addition, the calibration curve was used to evaluate the calibration of the model, and decision curve analysis (DCA) was used to assess the clinical utility of the model. All internal validations were conducted using the Bootstrap method, with a total of 1,000 resamples ( 16 , 17 ). This helped to balance the differences in sample distribution, thereby more accurately validating the generalization ability and robustness of the model.

Results

Patient characteristics

A total of 105 pediatric patients who met the inclusion criteria were enrolled in this study, comprising 56 males and 49 females ( Figure 1 ). The male-to-female incidence ratio was approximately 1.14:1. The median age at onset was 55 months (IQR: 29, 77). Among these patients, 16 (15.24%) presented with disease onset before 18 months of age, while 89 (84.76%) had an onset at or after 18 months of age. According to the INPC system, postoperative histopathological analysis revealed neuroblastoma (NB) in 29 cases, ganglioneuroblastoma nodular (GNBn) in 7 cases, ganglioneuroblastoma intermixed (GNBi) in 49 cases, and ganglioneuroma (GN) in 20 cases. Based on the INSS, 30 patients (28.57%) were classified as stage 1, 53 (50.48%) as stage 2, 12 (11.43%) as stage 3, and 10 (9.52%) as stage 4. No patient was identified as stage 4S in the study cohort. Baseline characteristics and relevant clinical data are summarized in Table 1 .

Among the 105 patients with thoracic neuroblastoma enrolled in the study, 64 (60.95%) exhibited a total of 93 IDRFs based on the final preoperative CT or MRI examination. Of these 64 patients, 44 had infiltrative IDRFs, 16 had vascular IDRFs, 9 had neurological IDRFs, 19 had extensive IDRFs, and 7 presented with preoperative pleural effusion. With regard to the distribution of IDRF counts, 42 patients (65.63%) had 1 IDRF, 16 patients (25.00%) had 2 IDRFs, 5 patients (7.81%) had 3 IDRFs, and 1 patient (1.56%) had 4 IDRFs. The proportion of IDRF counts is illustrated in Supplementary Figure 1 .

Analysis of relevant factors influencing IDRFs

To investigate potential factors influencing the occurrence of IDRFs, the logistic regression analysis was performed using demographic, clinical, and pathological characteristics of 105 pediatric patients with neuroblastoma as independent variables, with the presence or absence of IDRFs serving as the dependent variable. Before the analysis, the R package “Cutoff” was employed to determine the optimal cutoff values for NSE, TP, and MTD, which were identified as 30.27 ng/mL, 69.90 g/L, and 5.50 cm, respectively. LDH, HDL, and Hb levels were categorized based on their median values of 249.00 U/L, 1.29 mmol/L, and 125.00 g/L, respectively, while age was stratified at 18 months. Initially, univariate logistic regression analysis was conducted on the aforementioned variables. The results indicated that NSE ≥30.27 ng/mL ( p = 0.007, OR = 4.85, 95% CI: 1.53–15.35), TP ≥ 69.90 g/L ( p = 0.001, OR = 8.12, 95% CI: 2.26–29.15), and MTD ≥5.50 cm ( p < 0.001, OR = 5.93, 95% CI: 2.42–14.52) were significantly associated with the occurrence of IDRFs. Variables with p < 0.10 were subsequently included in a multivariate logistic stepwise regression analysis, which identified TP level and MTD as independent risk factors for IDRFs. Patients with TP ≥ 69.90 g/L exhibited 6.90-fold increased risk of IDRFs compared to those with TP < 69.90 g/L, while patients with MTD ≥5.50 cm had 5.09-fold higher risk compared to those with a diameter <5.50 cm ( Figure 2 ).

Correlation analysis between IDRFs and prognosis of thoracic neuroblastoma

Survival characteristics

Surgical therapy serves as the cornerstone in the management of localized neuroblastoma ( 10 , 11 ). Thus, thoracoscopy was the preferred initial approach, enabling complete tumor resection and involved lymph node dissection. For paravertebral tumors with intraspinal invasion, large tumor volume, severe adhesion of the tumor to surrounding tissues and organs, or difficulty in tumor resection under thoracoscopy, thoracotomy was adopted. After excluding 20 patients who were postoperatively diagnosed with ganglioneuroma based on pathological examination, a total of 85 pediatric patients were enrolled for survival analysis. To evaluate the prognostic significance of IDRFs in neuroblastoma, the Kaplan–Meier method was employed to assess the association between the presence or absence of IDRFs and both overall survival (OS) and event-free survival (EFS). The follow-up period extended until December 2024. During the follow-up, 2 patients died (both confirmed as neuroblastoma by postoperative pathology), and 25 patients experienced tumor recurrence, metastasis, or the development of secondary malignancies (including 10 cases of GNBi, 2 cases of GNBn, and 13 cases of NB). According to the Kaplan–Meier survival analysis, the 5-year overall survival rate was 100% in the no-IDRFs group and 96.1% in the IDRFs group. However, no statistically significant difference in OS was observed between the two groups ( p = 0.289, Figure 3A ). For event-free survival, the 1-year EFS rate was 96.7%, the 3-year EFS rate was 90.2%, and the 5-year EFS rate was 77.3% in the no-IDRFs group, compared to 74.9%, 51.0%, and 51.0%, respectively, in the IDRFs group. A statistically significant difference in EFS was observed between the two groups ( p < 0.05), as shown in Figure 3B .

Analysis of factors associated with event-free survival (EFS)

Different types of IDRFs (infiltrative, vascular, neurological, and extensive) as well as varying numbers of IDRFs were incorporated into the analysis. The optimal cutoff values for NSE, LDH, TP, and MTD were determined using the R package “Cutoff”, yielding values of 34.21 ng/mL, 273.00 U/L, 70.00 g/L, and 4.80 cm, respectively. Age was categorized at 18 months, while HDL and Hb levels were stratified based on their median values of 1.28 mmol/L and 124.00 g/L, respectively. Using the presence or absence of adverse prognostic events as the grouping variable, the aforementioned indicators were compared and analyzed. Statistically significant differences between the two groups were observed in terms of EFS, INPC pathological classification, infiltrative IDRFs, vascular IDRFs, extensive IDRFs, number of IDRFs, NSE levels, LDH levels, TP levels, and MTD ( p < 0.05), as summarized in Table 2 .

Subsequently, both univariate and multivariate Cox regression analyses were performed to evaluate the predictive value of these factors on event-free survival (EFS) in patients with neuroblastoma. Univariate Cox regression analysis revealed that INPC pathology classified as neuroblastoma ( p = 0.010, HR = 2.87, 95% CI: 1.28–6.42), presence of infiltrative IDRFs ( p = 0.038, HR = 2.29, 95% CI: 1.05–5.00), vascular IDRFs ( p < 0.001, HR = 5.60, 95% CI: 2.53–12.40), extensive IDRFs ( p = 0.006, HR = 2.94, 95% CI: 1.36–6.36), presence of two IDRFs ( p = 0.008, HR = 6.12, 95% CI: 1.62–23.12), presence of three or more IDRFs ( p < 0.001, HR = 17.37, 95% CI: 4.24–71.20), NSE ≥ 34.21 ng/mL ( p < 0.001, HR = 4.54, 95% CI: 2.07–9.95), LDH ≥ 273.00 U/L ( p = 0.003, HR = 3.27, 95% CI: 1.50–7.10), TP ≥ 70.00 g/L ( p = 0.009, HR = 2.81, 95% CI: 1.30–6.10), and MTD ≥ 4.80 cm ( p = 0.009, HR = 3.33, 95% CI: 1.34–8.27) were significantly associated with EFS in patients with neuroblastoma. All variables with p < 0.10 in the univariate analysis were included in the multivariate Cox regression. The optimal model variables were selected using bidirectional stepwise regression based on the AIC, identifying infiltrative IDRFs, vascular IDRFs, LDH ≥273.00 U/L, and TP ≥ 70.00 g/L as independent prognostic factors influencing EFS ( Table 3 ). The multicollinearity analysis was conducted for these four variables, and their variance inflation factor (VIF) values were calculated as 1.885, 1.745, 1.235, and 1.175, respectively, all below 5, indicating no significant multicollinearity among the selected factors.

Development of nomogram for predicting EFS

The predictive nomogram model for EFS was developed based on four independent prognostic factors identified through univariate and multivariate Cox regression analyses. In this model, the presence of infiltrative IDRFs was assigned a score of 46.67, while their absence received a score of 0; the presence of vascular IDRFs was assigned 100 points, and absence received 0. For LDH levels ≥273.00 U/L, a score of 66.67 was assigned, whereas levels below this threshold received 0. Similarly, TP levels ≥70.00 g/L were assigned 71.11 points, and lower levels received 0. Using the nomogram, the total score from one or more prognostic factors can be calculated to estimate the 1-year, 3-year, and 5-year EFS rates in thoracic neuroblastoma. The lower total score indicates a reduced risk of adverse prognostic events, whereas a higher score corresponds to an increased risk ( Figure 4 ).

Validation and performance of the nomogram

To evaluate the discriminatory ability of the model, we calculated the concordance index (C-index) and time-dependent AUC curves. The results indicated that the overall C-index of the prognostic model was 0.77 (95% CI: 0.69–0.86), which was higher than 0.70. Additionally, the time-dependent AUC curves demonstrated that, compared with any single prognostic marker, the model exhibited superior prognostic accuracy in predicting 1-year, 3-year, and 5-year EFS rates in tumor patients ( Figure 5A ). Similar results were also observed when internal validation was conducted by resampling 1,000 times using the Bootstrap method ( Figure 5B ).

Calibration curves were constructed to assess the calibration of the model in predicting 1-year, 3-year, and 5-year EFS rates. The results revealed a high degree of overlap between the predicted survival curves (red solid lines) and the ideal reference curves (black dashed lines) for 1-year, 3-year, and 5-year EFS, indicating a strong agreement between the model-predicted and observed survival probabilities ( Figure 6 ).

The clinical utility of the prediction model was further evaluated using DCA curves. In DCA curves, the x -axis represents the threshold probability, and the y -axis represents the net benefit. The green horizontal line and the blue diagonal line represent two extreme scenarios: the former assumes that no patients have risk factors, while the latter assumes that all patients have all risk factors, thereby ignoring individual variability ( Figure 7 ). The results showed that the model achieved a favorable net benefit when the threshold probabilities for predicting 1-year, 3-year, and 5-year EFS were within the ranges of approximately 0.02–0.73, 0.08–0.84, and 0.09–0.85, respectively. These findings indicate that the model has potential clinical applicability.

Discussion

Thoracic neuroblastoma predominantly arises from the sympathetic ganglion regions within the posterior mediastinum, particularly the paravertebral groove, with an estimated incidence of 15%–20% ( 18 , 19 ). Previous studies have demonstrated that neuroblastomas originating from different anatomical sites display marked heterogeneity in clinical features, biological behavior, and prognostic outcomes. Therefore, the present study retrospectively collected clinical and prognostic data from a single-center cohort of pediatric patients with thoracic neuroblastoma, with a specific focus on the distribution of IDRFs and their prognostic significance. The median age at diagnosis in this cohort was 55 months, with 89 patients (84.76%) diagnosed at or above 18 months of age. In comparison, previous studies have reported a median diagnostic age of approximately 2 years old ( 20 ), with nearly 40% of neuroblastoma cases diagnosed before the age of 1 year ( 18 , 21 ). Furthermore, approximately 50% of thoracic neuroblastoma cases typically present at ≥18 months of age ( 9 , 22 ). These comparisons suggested that the current cohort represented a relatively older diagnostic age group. Regarding histopathological classification, GNBi was the most frequently observed subtype, accounting for 46.67%, followed by classical NB at 19.05%. The observed findings align with the pathological distribution patterns reported by Yang S et al. ( 23 ), yet they contrasted with the results presented by Demir et al. ( 24 ) and Butzer et al. ( 9 ), which may indicate potential variations in the biological behavior of thoracic neuroblastoma across different populations and geographic regions. In contrast, the distribution of INSS staging and MYCN gene amplification in this study was consistent with previously published data.

Imaging-defined risk factors (IDRFs) evolved from the surgical risk factors (SRFs) previously established by the International Society of Pediatric Oncology-European Neuroblastoma Group (SIOPEN) to evaluate the feasibility and safety of surgical resection ( 25 ). Given their critical role in predicting surgical outcomes and clinical prognosis, IDRFs have garnered increasing attention in recent research. In this study, the classification system proposed by Delforge et al. ( 13 ) was applied to categorize thoracic neuroblastoma-related IDRFs into four major categories: infiltrative, vascular, neurological, and extensive. Among the 105 patients in the study cohort, 64 (60.95%) were identified as having at least one IDRF based on the final preoperative imaging evaluation. Regarding the number of IDRFs, the majority of patients exhibited a single IDRF (42 cases, 65.63%). In terms of type distribution, infiltrative IDRFs were the most frequently observed (44 cases, 68.75%), followed by vascular IDRFs (19 cases, 29.69%). To further investigate potential factors associated with the presence of IDRFs, demographic characteristics, pathological findings, preoperative laboratory tests, and imaging results were analyzed. Univariate and multivariate logistic regression analyses identified serum TP ≥ 69.90 g/L and MTD ≥ 5.50 cm as independent predictors of IDRF presence. Previous studies examining factors associated with IDRFs ( 26 – 28 ) have consistently highlighted tumor size as a key influencing factor, a finding corroborated in this study. Specifically, patients with tumors size ≥5.50 cm exhibited 5.09-fold increased likelihood of harboring IDRFs compared to those with smaller tumors. However, variations in patient age, sample size, clinical staging, and imaging modalities across studies led to discrepancies in the proposed tumor size thresholds. Therefore, future multicenter studies with larger cohorts are warranted to further validate these findings.

Another critical factor influencing the presence of IDRFs was the level of serum TP in patients. As a key biochemical marker, TP reflects various physiological and pathological conditions, including nutritional status, immune function, and inflammatory responses ( 29 – 31 ). It plays a significant role in evaluating disease severity, hepatic and renal function, and clinical prognosis. Moreover, due to its well-established testing methodology, cost-effectiveness, and wide clinical applicability, TP measurement is routinely performed in clinical settings. Previous studies have indicated that during carcinogenesis, elevated oxidative stress levels combined with impaired antioxidant defenses may lead to alterations in serum protein concentrations. More et al. ( 32 ) reported that increased serum protein levels may be associated with inflammatory responses in oral malignant tumors. Specifically, globulin, a major component of TP, was found to be significantly elevated in patients with oral cancer, suggesting that this increase may be attributed to its role as an acute-phase reactant. Similarly, Nimbal et al. ( 33 ) observed significantly higher TP levels in patients with smokeless tobacco-induced oral malignancies compared to those with precancerous lesions. Ghalehsari et al. ( 34 ) also reported elevated TP levels in patients with multiple myeloma, further supporting the association between TP and malignancy. Recent studies demonstrated that the neuroblastoma microenvironment contains a substantial number of immune and stromal cells, including M2 macrophages, natural killer T cells (NKT), tumor-infiltrating T lymphocytes (TIL-T), and cancer-associated fibroblasts (CAF) ( 35 – 37 ). These cells secreted various cytokines and proteases that suppress anti-tumor immunity, promote local inflammation, and facilitate tumor progression and metastasis. These findings suggested that TP levels might serve as a potential biomarker for neuroblastoma, reflecting disease severity. In our study, patients with TP ≥ 69.90 g/L exhibited 6.90-fold increased risk of harboring IDRFs compared to those with TP < 69.90 g/L. This association might be attributed to TP's capacity to reflect tumor burden and the degree of malignancy. Higher TP levels were indicative of more aggressive tumor biology, which in turn increases the probability of IDRFs occurrence.

The prognosis of patients with neuroblastoma has long been a central focus for clinical researchers. With the development and clinical application of novel therapeutic strategies, including immunotherapy, stem cell transplantation, and CAR-T cell therapy ( 38 – 40 ), the 5-year OS for patients with neuroblastoma has shown improvement over the past three decades ( 41 ). A recent population-based study reported that the 5-year survival rate for pediatric patients with neuroblastoma in the United States reached 79.7% by 2019 ( 42 ). Nevertheless, existing literature suggested that even among postoperative survivors, long-term outcomes and quality of life are not optimistic ( 43 ). In the present study, the 5-year OS was 100% in the group without imaging-defined risk factors (IDRFs) and 96.1% in the group with IDRFs. These results were consistent with previously published findings ( 44 – 46 ), yet higher than the survival rates reported for patients with high-risk neuroblastoma by Espinoza et al. ( 47 ). This suggested that thoracic neuroblastomas might generally carry a more favorable prognosis. However, a statistically significant difference in EFS was observed between the two groups ( p < 0.05). Through group difference analysis and multivariate Cox regression analysis based on the Akaike information criterion (AIC), infiltrative IDRFs, vascular IDRFs, LDH ≥273.00 U/L, and TP ≥ 70.00 g/L were identified as independent prognostic factors influencing EFS in pediatric patients.

Numerous studies have demonstrated the significant prognostic value of imaging-defined risk factors (IDRFs) in neuroblastoma. A meta-analysis conducted by Parhar et al. revealed that patients with positive IDRFs exhibited significantly lower five-year OS and EFS compared to those without IDRFs ( 48 ). This finding was further supported by the most recent study from Yang et al. ( 28 ). Moreover, emerging evidence suggested that the presence of IDRFs may be associated with specific clinical and biological features of neuroblastoma. Factors such as larger tumor volume, MYCN gene amplification, and elevated serum levels of NSE and LDH have been implicated in this association ( 27 , 28 , 41 ). Notably, recent investigations indicated that not all IDRF subtypes demonstrated equivalent prognostic significance. Features such as major vascular encasement and intraspinal tumor infiltration were often indicative of highly aggressive tumor behavior and were associated with a poorer prognosis ( 13 , 49 ).

For focal neuroblastoma, surgery is the cornerstone of treatment ( 10 , 11 ). There have been numerous previous studies demonstrated that the presence of IDRFs was consistently associated with increased surgical complexity, prolonged postoperative hospitalization, reduced rates of gross-total resection, and higher incidence of perioperative complications ( 47 , 50 – 53 ). Especially for tumors with vascular IDRFs, the tumor volume was often larger, and the difficulty of separating the tumor from the vessels was higher, which undoubtedly brought extremely high difficulty and risk to thoracoscopic surgery and increases the rate of incomplete tumor resection ( 44 , 53 ). Saksiri et al. reported that there was a significant difference in the occurrence of surgical complications between the groups with or without IDRFs, with approximately over 60% of neuroblastoma patients with IDRFs experiencing surgery-related complications ( 52 ). These factors undoubtedly worsen the prognosis of the patients and have a significant impact on the treatment outcome.

In this study, we first integrated IDRFs into the development of a nomogram designed to predict EFS in patients with thoracic neuroblastoma. The calibration curves for 1-year, 3-year, and 5-year EFS demonstrated excellent agreement with the ideal reference curves, indicating good calibration and prediction accuracy. Furthermore, the overall C-index of the prognostic model exceeded 0.70, and DCA curves revealed that the model provided a favorable net benefit across a broad range of threshold probabilities. Therefore, this nomogram might serve as a valuable tool for estimating EFS in patients with thoracic neuroblastoma, assisting clinicians in developing more individualized and precise treatment strategies, thereby potentially improving long-term survival and quality of life in affected children.

Certainly, this study also has some limitations. First, the low prevalence of MYCN amplification in this cohort precluded its inclusion in the prognostic analysis. Moreover, due to economic cost and diagnostic technology limitations, other established prognostic biomarkers, such as segmental chromosomal aberrations (SCAs) and anaplastic lymphoma kinase (ALK) status, were not evaluated, resulting in the absence of relevant data. This has limited the reliability and generalizability of the prognostic analysis in this study. Second, the sample size of this study was limited and all data are from a single center, which might lead to potential selection bias. Therefore, future multi-center studies with larger sample sizes are warranted to validate our findings. Finally, the current prediction model was specifically developed for thoracic neuroblastoma and might not be directly applicable to tumors originating from other anatomical sites. So, further investigations are warranted to assess its performance in anatomical locations.

Conclusions

Thoracic neuroblastoma exhibited distinct demographic, clinical, and pathological characteristics compared to nonthoracic tumors. TP level and maximum tumor diameter were independently associated with the presence of IDRFs. IDRFs significantly fluenced EFS in pediatric patients with thoracic neuroblastoma, with infiltrative IDRFs, vascular IDRFs, elevated LDH, and increased TP levels identified as independent prognostic risk factors for EFS. The nomogram model developed based on these risk factors demonstrated robust predictive performance and held potential for clinical application.

Acknowledgments

This study has been approved by the Hospital Ethics Committee of the Children's Hospital of Nanjing Medical University (Permission No.202205047-1). All procedures involving human participants in this study were in accordance with the ethical standards of the institutional research committee and the Helsinki Declaration.

Funding Statement

The author(s) declared that financial support was not received for this work and/or its publication.

Footnotes

Edited by: Erin Brown , University of California, Davis, United States

Reviewed by: Hany Gabra , Newcastle University, United Kingdom

Stefano Avanzini , Giannina Gaslini Institute (IRCCS), Italy

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

The studies involving humans were approved by Hospital Ethics Committee of the Children's Hospital of Nanjing Medical University (Permission No. 202205047-1). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants' legal guardians/next of kin in accordance with the national legislation and institutional requirements.

Author contributions

CS: Supervision, Conceptualization, Writing – original draft, Data curation, Software, Visualization, Formal analysis, Methodology. ZB: Visualization, Validation, Data curation, Methodology, Writing – review & editing, Investigation. KZ: Investigation, Visualization, Data curation, Writing – review & editing, Formal analysis. YM: Writing – review & editing, Investigation, Data curation, Visualization, Formal analysis. MY: Investigation, Supervision, Data curation, Writing – review & editing, Validation. JQ: Data curation, Investigation, Writing – review & editing, Validation, Supervision. BQ: Data curation, Conceptualization, Project administration, Validation, Writing – review & editing, Supervision, Formal analysis. KW: Writing – review & editing, Validation, Project administration, Formal analysis, Methodology, Conceptualization, Supervision.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fped.2026.1793866/full#supplementary-material

References

1. Shohet J, Foster J. Neuroblastoma. Br Med J. (2017) 357:j1863. 10.1136/bmj.j1863

2. Tas ML, Reedijk AMJ, Karim-Kos HE, Kremer LCM, van de Ven CP, Dierselhuis MP, et al. Neuroblastoma between 1990 and 2014 in The Netherlands: increased incidence and improved survival of high-risk neuroblastoma. Eur J Cancer. (2020) 124:47–55. 10.1016/j.ejca.2019.09.025

3. Tsubota S, Kadomatsu K. Origin and initiation mechanisms of neuroblastoma. Cell Tissue Res. (2018) 372:211–21. 10.1007/s00441-018-2796-z

4. Park JR, Bagatell R, Cohn SL, Pearson AD, Villablanca JG, Berthold F, et al. Revisions to the international neuroblastoma response criteria: a consensus statement from the national cancer institute clinical trials planning meeting. J Clin Oncol. (2017) 35:2580–7. 10.1200/jco.2016.72.0177

5. He M, Cai JB, Wu X, Tang YB, Wang JY, Mao JQ, et al. Perioperative complication incidence and risk factors for retroperitoneal neuroblastoma in children: analysis of 571 patients. World J Pediatr. (2024) 20:250–8. 10.1007/s12519-023-00773-z

6. Wang H. Reflections on surgical treatment of neuroblastoma. J Clin Ped Sur. (2022) 21:101–3. 10.3760/ama.j.cn.101785-202202032-001

7. Monclair T, Brodeur GM, Ambros PF, Brisse HJ, Cecchetto G, Holmes K, et al. The international neuroblastoma risk group (INRG) staging system: an INRG task force report. J Clin Oncol. (2009) 27:298–303. 10.1200/jco.2008.16.6876

8. Spencer B, Patel A, Cilley R, Grant CN. Surgical management in pediatric neuroblastoma diagnosis and treatment: a 20-year, single-center experience. World J Pediatr. (2022) 18:120–5. 10.1007/s12519-021-00490-5

9. Butzer SK, Krug B, Reisberg M, Fuchs J, Warmann SW, Hubertus J, et al. Image-defined risk factors in localized thoracic neuroblastoma and ganglioneuroma. Pediatr Blood Cancer. (2023) 70:e30605. 10.1002/pbc.30605

10. Croteau N, Nuchtern J, LaQuaglia MP. Management of neuroblastoma in pediatric patients. Surg Oncol Clin N Am. (2021) 30:291–304. 10.1016/j.soc.2020.11.010

11. Ryan AL, Akinkuotu A, Pierro A, Morgenstern DA, Irwin MS. The role of surgery in high-risk neuroblastoma. J Pediatr Hematol Oncol. (2020) 42:1–7. 10.1097/mph.0000000000001607

12. Peuchmaur M, d'Amore ES, Joshi VV, Hata J, Roald B, Dehner LP, et al. Revision of the international neuroblastoma pathology classification: confirmation of favorable and unfavorable prognostic subsets in ganglioneuroblastoma, nodular. Cancer. (2003) 98:2274–81. 10.1002/cncr.11773

13. Delforge X, De Cambourg P, Defachelles AS, Haffreingue A, Rod J, Kassite I, et al. Unresectable thoracic neuroblastic tumors: changes in image-defined risk factors after chemotherapy and impact on surgical management. Pediatr Blood Cancer. (2021) 68:e29260. 10.1002/pbc.29260

14. Iasonos A, Schrag D, Raj GV, Panageas KS. How to build and interpret a nomogram for cancer prognosis. J Clin Oncol. (2008) 26:1364–70. 10.1200/jco.2007.12.9791

15. Graesslin O, Abdulkarim BS, Coutant C, Huguet F, Gabos Z, Hsu L, et al. Nomogram to predict subsequent brain metastasis in patients with metastatic breast cancer. J Clin Oncol. (2010) 28:2032–7. 10.1200/jco.2009.24.6314

16. Van Calster B, Nieboer D, Vergouwe Y, De Cock B, Pencina MJ, Steyerberg EW. A calibration hierarchy for risk models was defined: from utopia to empirical data. J Clin Epidemiol. (2016) 74:167–76. 10.1016/j.jclinepi.2015.12.005

17. Zong F, Zhang X, Hua R, Zhang S, Qiao H, Duan Y, et al. A novel prognostic model integrating host vulnerability and tumor biology for elderly patients with diffuse large B-cell lymphoma: the BAMAL score. Front Oncol. (2026) 16:1720879. 10.3389/fonc.2026.1720879

18. Matthay KK, Maris JM, Schleiermacher G, Nakagawara A, Mackall CL, Diller L, et al. Neuroblastoma. Nat Rev Dis Primers. (2016) 2:16078. 10.1038/nrdp.2016.78

19. Swift CC, Eklund MJ, Kraveka JM, Alazraki AL. Updates in diagnosis, management, and treatment of neuroblastoma. Radiographics. (2018) 38:566–80. 10.1148/rg.2018170132

20. Fisher JPH, Tweddle DA. Neonatal neuroblastoma. Semin Fetal Neonatal Med. (2012) 17:207–15. 10.1016/j.siny.2012.05.002

21. Johnsen JI, Dyberg C, Wickström M. Neuroblastoma-A neural crest derived embryonal malignancy. Front Mol Neurosci. (2019) 12:9. 10.3389/fnmol.2019.00009

22. Oldridge DA, Truong B, Russ D, DuBois SG, Vaksman Z, Mosse YP, et al. Differences in genomic profiles and outcomes between thoracic and adrenal neuroblastoma. J Natl Cancer Inst. (2019) 111:1192–201. 10.1093/jnci/djz027

23. Yang S, Cai S, Wang H, Zeng Q, Ma X, Peng X, et al. Clinical and prognostic characteristics of neuroblastoma in children with abdominal and thoracic primary tumor sites. Chin J Pediatr Surg. (2019) 40:488–93. 10.3760/cma.j.issn.0253-3006.2019.06.002

24. Demir HA, Yalçin B, Büyükpamukçu N, Kale G, Varan A, Akyüz C, et al. Thoracic neuroblastic tumors in childhood. Pediatr Blood Cancer. (2010) 54:885–9. 10.1002/pbc.22405

25. Cecchetto G, Mosseri V, De Bernardi B, Helardot P, Monclair T, Costa E, et al. Surgical risk factors in primary surgery for localized neuroblastoma: the LNESG1 study of the European international society of pediatric oncology neuroblastoma group. J Clin Oncol. (2005) 23:8483–9. 10.1200/jco.2005.02.4661

26. Phelps HM, Ndolo JM, Van Arendonk KJ, Chen H, Dietrich HL, Watson KD, et al. Association between image-defined risk factors and neuroblastoma outcomes. J Pediatr Surg. (2019) 54:1184–91. 10.1016/j.jpedsurg.2019.02.040

27. Laura B, Radu B, Patricia C, Andreea M. Association of image-defined risk factors with clinical, biological features and outcome in neuroblastoma. Children (Basel). (2022) 9:1707. 10.3390/children9111707

28. Yang DD, Liu C, Gao J, Hu QJ, Liang Y, Liu J. Association of image-defined risk factors with clinical features, tumor biology, and outcomes in neuroblastoma: a single-center retrospective study. Eur J Pediatr. (2023) 182:2189–96. 10.1007/s00431-023-04899-0

29. Inoue M, Takeuchi J, Sakuma M, Nakamura T, Morimoto T. Low Serum total protein at admission predicts in-hospital mortality among general inpatients: historical cohort study. Int J Gen Med. (2022) 15:7941–9. 10.2147/ijgm.S385798

30. Yin T, Wei W, Huang X, Liu C, Li J, Yi C, et al. Serum total protein-to-albumin ratio predicts risk of death in septic acute kidney injury patients: a cohort study. Int Immunopharmacol. (2024) 127:111358. 10.1016/j.intimp.2023.111358

31. Dawnay AB, Hirst AD, Perry DE, Chambers RE. A critical assessment of current analytical methods for the routine assay of serum total protein and recommendations for their improvement. Ann Clin Biochem. (1991) 28(Pt 6):556–67. 10.1177/000456329102800604

32. More CB, Shah PH, Venkatesh R. Estimation of Serum protein in oral potentially malignant disorders and oral malignancy—a cross-sectional study. J Clin Diagn Res. (2017) 11:Zc17-zc9. 10.7860/jcdr/2017/20815.9254

33. Nimbal A, Ahirrao B, Vishwakarma A, Vishwakarma P, Wani AB, Patil AA. Comparative evaluation of GSH, total protein and albumin levels in patients using smokeless tobacco with oral precancerous and cancerous lesions. Med Int (Lond). (2024) 4:15. 10.3892/mi.2024.139

34. Ghalehsari N, Zahid S, Main O, Usta S, Patresan J, Amin A, et al. Exploring health disparities in diagnosing multiple myeloma. Expert Rev Hematol. (2024) 17:749–53. 10.1080/17474086.2024.2389988

35. Qiu B, Matthay KK. Advancing therapy for neuroblastoma. Nat Rev Clin Oncol. (2022) 19:515–33. 10.1038/s41571-022-00643-z

36. Li S, Luo J, Liu J, He D. Pan-cancer single cell and spatial transcriptomics analysis deciphers the molecular landscapes of senescence related cancer-associated fibroblasts and reveals its predictive value in neuroblastoma via integrated multi-omics analysis and machine learning. Front Immunol. (2024) 15:1506256. 10.3389/fimmu.2024.1506256

37. Stip MC, Teeuwen L, Dierselhuis MP, Leusen JHW, Krijgsman D. Targeting the myeloid microenvironment in neuroblastoma. J Exp Clin Cancer Res. (2023) 42:337. 10.1186/s13046-023-02913-9

38. Prete A, Lanino E, Saglio F, Biffi A, Calore E, Faraci M, et al. Phase II study of allogeneic hematopoietic stem cell transplantation for children with high-risk neuroblastoma using a reduced-intensity conditioning regimen: results from the AIEOP trial. Transplant Cell Ther. (2024) 30:530.e1–e8. 10.1016/j.jtct.2024.03.002

39. Sun M, Cao Y, Okada R, Reyes-González JM, Stack HG, Qin H, et al. Preclinical optimization of a GPC2-targeting CAR T-cell therapy for neuroblastoma. J Immunother Cancer. (2023) 11:e005881. 10.1136/jitc-2022-005881

40. Boldajipour B, Lee G, Albertson T. GD2-CART01 For relapsed or refractory high-risk neuroblastoma. N Engl J Med. (2023) 388:2303. 10.1056/NEJMc2305296

41. Wang H, Mejia MC, Gonzalez SJ, Zoorob RJ, Chai W, Du XL. Cancer incidence and survival trends among infants in the United States from 1975 to 2014. Pediatr Blood Cancer. (2021) 68:e28917. 10.1002/pbc.28917

42. Campbell K, Siegel DA, Umaretiya PJ, Dai S, Heczey A, Lupo PJ, et al. A comprehensive analysis of neuroblastoma incidence, survival, and racial and ethnic disparities from 2001 to 2019. Pediatr Blood Cancer. (2024) 71:e30732. 10.1002/pbc.30732

43. Friedman DN, Goodman PJ, Leisenring WM, Diller LR, Cohn SL, Howell RM, et al. Long-Term morbidity and mortality among survivors of neuroblastoma diagnosed during infancy: a report from the childhood cancer survivor study. J Clin Oncol. (2023) 41:1565–76. 10.1200/jco.22.01732

44. Lecompte JF, Sarnacki S, Irtan S, Piolat C, Scalabre A, Talon I, et al. Thoracoscopy for pediatric thoracic neurogenic tumors-A European multi-center study. Cancers (Basel). (2023) 15:5467. 10.3390/cancers15225467

45. Rossa A, Cacciavillano W, Rose A, Flores P, Chantada G, Zubizarreta P. Neuroblastoma in patients under 18 months. Single institution experience in Argentina. Medicina (B Aires). (2019) 79:280–3. PMID: .

46. Simon T, Hero B, Benz-Bohm G, von Schweinitz D, Berthold F. Review of image defined risk factors in localized neuroblastoma patients: results of the GPOH NB97 trial. Pediatr Blood Cancer. (2008) 50:965–9. 10.1002/pbc.21343

47. Espinoza AF, Bagatell R, McHugh K, Naranjo AH, Van Ryn C, Rojas Y, et al. A subset of image-defined risk factors predict completeness of resection in children with high-risk neuroblastoma: an international multicenter study. Pediatr Blood Cancer. (2024) 71:e31218. 10.1002/pbc.31218

48. Parhar D, Joharifard S, Lo AC, Schlosser MP, Daodu OO. How well do image-defined risk factors (IDRFs) predict surgical outcomes and survival in patients with neuroblastoma? A systematic review and meta-analysis. Pediatr Surg Int. (2020) 36:897–907. 10.1007/s00383-020-04696-y

49. Zhang AA, Pan C, Xu M, Wang XX, Ye QD, Gao YJ, et al. Association of image-defined risk factors, tumor resectability, and prognosis in children with localized neuroblastoma. World J Pediatr. (2019) 15:572–9. 10.1007/s12519-019-00274-y

50. Chang S, Yang S, Yu T, Zhang Q, Lin Y, Ren Q, et al. A primary report 166 cases of abdominal or pelvic neuroblastoma surgery utilizing the international neuroblastoma surgical report form (INSRF). Ann Surg. (2025) 282:132–7. 10.1097/sla.0000000000006292

51. Chang S, Lin Y, Yang S, Yang W, Cheng H, Chang X, et al. Safety and feasibility of laparoscopic resection of abdominal neuroblastoma without image-defined risk factors: a single-center experience. World J Surg Oncol. (2023) 21:113. 10.1186/s12957-023-02997-9

52. Saksiri P, Sripornsawan P, Kritsaneepaiboon S, Laochareonsuk W, Choochuen P, Sangkhathat S. Correlation between image-defined risk factors and surgical complications in patients with neuroblastoma: a retrospective study. Pediatr Surg Int. (2023) 39:70. 10.1007/s00383-022-05361-2

53. Froeba-Pohl A, von Schweinitz D, Muehling J, Paolini M, Hubertus J. Implication of image-defined risk factors for the extent of surgical resection and clinical outcome in patients with pelvic neuroblastoma. Eur J Pediatr Surg. (2021) 31:362–6. 10.1055/s-0040-1714658

Associated Data

Supplementary Materials

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.