Objective  To summarize and analyze serum neuron-specific enolase (NSE) and urine vanillylmandelic acid (VMA) levels in children with neuroblastoma before treatment by correlating these levels with clinicopathological features of neuroblastomas, establishing their ability to predict high-risk children, and finally, providing a basis for better diagnosis and assessment of neuroblastomas.

  Methods  Clinicopathological data of 155 children with a confirmed diagnosis of neuroblastoma at Tianjin Medical University Cancer Institute and Hospital from September 30, 2011 to September 11, 2019 were collected. A retrospective analysis was performed to analyze serum NSE and urine VMA levels in children with different clinical features such as age, gender, tumor size and location, pathological pattern, staging, risk grouping, metastasis, MYCN gene amplification, and image-defined risk factors (IDRFs). Differences between groups were compared using a nonparametric rank-sum test. A P-value < 0.05 was considered statistically significant. Receiver operating characteristic (ROC) curves and prediction models were used to determine the accuracy of serum NSE and urine VMA in identifying high-risk children, thereby establishing the diagnostic value of these indicators.

  Results  Of 109 children tested for urine VMA, 61 and 48 children had negative and positive results, respectively, resulting in a positive rate of 44.0%. Of 152 children tested for serum NSE, 2 and 150 children had negative and positive results, respectively, suggesting a positive rate of 98.7%. Serum NSE and urine VMA levels were significantly increased in children with an INSS stage 4 classification, a COG high-risk grouping, metastases, and with presence of IDRFs (P < 0.05). The serum NSE level was also significantly different in children depending on the varying tumor sites, gross tumor volumes, and pathological patterns. Specifically, the NSE level was significantly higher in children with tumors >500 cm³, an unfavorable prognosis based on the INPC, a pathological pattern of neuroblastoma, and a retroperitoneal primary tumor focus than in children of other groups (P < 0.001). The ROC curves revealed that a serum NSE cut-off level of 98.50 ng/mL with an area under the curve (AUC) of 0.943 (95% confidence intervalCI, 0.906-0.991) and a urine VMA cut-off level of 62.28 μmol/24 h with an AUC of 0.791 (95% CI, 0.705-0.878) could identify high-risk children. When the two indicators were combined to identify high-risk children, the AUC was 0.948 (95% CI, 0.900-0.997).

  Conclusions  Compared with urine VMA, serum NSE levels correlated with more clinical features of neuroblastomas; thus, serum NSE could better assist in the diagnosis and evaluation of the disease by providing more comprehensive results. While both serum NSE and urine VMA levels are significantly higher in high-risk children, the accuracy, sensitivity, and specificity of serum NSE are all higher than those of urine VMA in identifying high-risk children. A combination of the two indicators does not improve diagnostic accuracy.