Table 1.
Patient characteristics. INSS = International Neuroblastoma Staging System.
Table 2.
Urine catecholamine metabolite levels and serum NSE in patients with favorable histopathology (FH) and unfavorable histopathology (UH) according to the International Neuroblastoma Pathology Classification (INPC).
Concentrations are given as mean ± standard deviation (SD). Difference was tested by unpaired Mann-Whitney test including Bonferroni correction for multiple comparisons (p < 0.005*). MN = metanephrine, VMA = vanillylmandelic acid, A = adrenaline, NA = noradrenaline, HVA = homovanillic acid, DP = dopamine, NSE = neuron-specific enolase.
Fig 1.
ROC of TLCRR and serum NSE for prediction of unfavorable histopathology.
Receiver-operating-characteristic (ROC) is given for TLCRR (dashed line) and serum NSE (continuous line). AUC-ROC was 0.86 for TLCRR. Optimal cut-off determined by the Youden index was 2.0 (*), resulting in a sensitivity of 68% and a specificity of 100%. AUC-ROC was 0.81 for serum NSE. Optimal cut-off was 25.8 ng/ml (**) with a sensitivity of 74% and a specificity of 85%. AUC-ROC = area under the receiver-operating-characteristic, TLCRR = tumor-to-liver count-rate ratio, NSE = neuron-specific enolase.
Fig 2.
Results for (A) TLCRR and (B) serum NSE with corresponding histopathology for each patient (n = 47).
Patients with unfavorable histopathology (UH) are shown with black bars, and patients with favorable histopathology (FH) are shown with white bars, all of which are ordered in ascending magnitude. Cut-off values determined by ROC analysis are indicated by horizontal lines. ROC = receiver-operating-characteristic, TLCRR = tumor-to-liver count-rate ratio, NSE = neuron-specific enolase.
Table 3.
Accuracy of several imaging parameters and tumor markers for prediction of unfavorable histopathology was tested by ROC analysis using Bonferroni correction for multiple comparisons. Corresponding AUC-ROC, p value and 95% CI are given.
Fig 3.
True positive prediction of unfavorable histopathology (UH) by combined TLCRR/NSE criteria in a 14 year old boy.
T1-weighted axial MRI with (A) and without (C) contrast enhancement, 123I-MIBG SPECT (D) and fused SPECT/MRI (B) of the lower abdomen are shown. Tumor is depicted as contrast enhancing lesion in a paravertebral location (A, arrow) with infiltration of the L5 neuroforamen (A, double arrow). Tumor-to-liver count-rate ratio (TLCRR) was < 2.0 (D, false negative). Serum neuron-specific enolase (NSE) level was > 25.8 ng/ml (D, true positive). Combined analysis resulted in a true positive finding for UH. Undifferentiated neuroblastoma was confirmed by histopathology following surgery.
Fig 4.
True positive prediction of unfavorable histopathology (UH) by combined TLCRR/NSE criteria in a 4 month old girl.
T1-weighted axial MRI with (A) and without (C) contrast enhancement, 123I-MIBG SPECT (D) and fused SPECT/MRI (B) of the abdomen are shown. Images show a lesion with contrast enhancement (A, arrow) and infiltration of the L2 neuroforamen (A, double arrow). Tumor-to-liver count-rate ratio (TLCRR) was > 2.0 (D, true positive). Serum neuron-specific enolase (NSE) level was < 25.8 ng/ml (D, false negative). Combined analysis resulted in a true positive finding for UH. Poorly differentiated neuroblastoma was confirmed by histopathology following surgery.
Table 4.
Performance of TLCRR, NSE and combined analysis for prediction of unfavorable histopathology.
Sensitivity (SE), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), and accuracy (AC) are each given in percent using optimal cut-off as determined by ROC analysis. Relative risk (RR) was calculated. TLCRR = tumor-to-liver count-rate ratio, NSE = neuron-specific enolase, ROC = receiver-operating-characteristic.