Abstract:
PURPOSE : Low- and middle-income countries (LMICs) reported a higher median age at diagnosis of neuroblastoma (NB) compared to high-income countries. The aim was to determine if the optimal age at diagnosis, which maximizes the difference in overall survival between younger versus older patients in the South African population was similar to the internationally validated 18 months age cut-point.
METHODS : Four hundred sixty NB patients diagnosed between 2000 and 2016 were included. Receiver operating characteristic (ROC) curves were used to predict potential age cut-point values for overall survival in all risk group classifications. Risk ratios, sensitivity, specificity, and positive and negative predictive values at the specific cut-points were estimated with 95% confidence intervals, and time to mortality by age at the specific cut-points was shown with Kaplan-Meier curves and compared using log-rank tests.
RESULTS : The median age at diagnosis for the total cohort was 31.9 months (range 0.2-204.7). For high-risk (HR), intermediate-risk, low-risk, and very low-risk patients, the median age at diagnosis was, respectively, 36 months (range 0.4-204.7), 16.8 months (range 0.7-145.1), 14.2 months (range 2.0-143.5), and 8.7 months (range 0.2-75.6). The ROC curves for the total NB cohort (area under the curve [AUC] 0.696; P < .001) and HR (AUC 0.682; P < .001) were analyzed further. The optimal cut-point value for the total cohort was at 19.1 months (sensitivity 59%; specificity 78%). The HR cohort had potential cut-point values identified at 18.4 months age at diagnosis (sensitivity 45%; specificity 87%) and 31.1 months (sensitivity 67%; specificity 62%). The 19.1 months cut-point value in the total cohort and the 18.4 months cut-point value in HR were as useful in predicting overall survival as 18 months age at diagnosis.
CONCLUSION : The 18 months cut-point value appears to be the appropriate age for prognostic determination, despite the higher median age at diagnosis in South Africa.
