Circulating miRNA in serum of Wilms tumor patients

Abstract: Background: MiRNA signatures in human sera have been reported for various tumor diseases. Here we generated miRNA profiles analyzing 1205 mature miRNA transcripts of serum samples of Wilms tumor patients, taken prior and after chemotherapy according to SIOP protocol 2001. Using a feature subset selection filter approach we identified a minimal number of miRNAs with a maximum contribution for the classification between treated and untreated patients and between patients and controls. Results: Analyzing 1205 mature miRNAs, we separated controls and Wilms tumor patients prior chemotherapy with an accuracy of 0.81. We obtained a similar accuracy (0.82) for the separation between controls and sera of Wilms tumor patients after preoperative chemotherapy. We identified 23 miRNAs that were differentially expressed in both comparisons. Subset selection improved the overall classification accuracy between controls and Wilms tumor patients prior and after chemotherapy to 0.94 and 0.91, respectively. Subset selection also allowed separating between Wilms tumor patients prior and after chemotherapy with an accuracy of 0.98. Conclusion: Our analysis identified serum based miRNA signatures that allowed separating between controls, untreated Wilms tumor patients, and Wilms tumor patients after chemotherapy with high accuracy for each of these comparisons. Serum miRNA profile of 16 Wilms tumor patients before and after preoperative chemotherapy, 19 sera of children without malignant disease