Supplementary Materials and Methods, Figures 1 - 10, Tables 1 - 4 from Noninvasive Urinary Metabolomic Profiling Identifies Diagnostic and Prognostic Markers in Lung Cancer
PDF file - 1526KB, Supplementary Table 1 shows random forest analysis results for predictions of lung cancer status in the training set. Supplementary Table 2 shows associations with survival in the training set when the top four predictive metabolites are combined in all cases. Supplementary Table 3 shows associations with survival in the training set, stratified by self-reported race. Supplementary Table 4 shows intraclass correlation coefficients in the quantitated subset. Supplementary Figure 1 depicts workflow of the classification analysis. Supplementary Figure 2 depicts quality control assessment in the training set. Supplementary Figure 3 shows predictions of smoking status in the training set determined by random forest analysis and abundances of tobacco-related metabolites. Supplementary Figure 4 shows overlap of metabolites predictive of lung cancer status in the training set based on random forest analysis, stratified by gender, race and smoking status. Supplementary Figure 5 shows fragmentation patterns of top four predictive metabolites determined by tandem mass spectrometry. Supplementary Figure 6 depicts identification of creatine riboside by NMR. Supplementary Figure 7 shows diurnal effects on top four predictive metabolites. Supplementary Figure 8 shows top four predictive metabolite abundances stratified by smoking status. Supplementary Figure 9 shows Kaplan-Meier survival estimates in the training set depicted for the top four predictive metabolites in stages I-II and their combination. Supplementary Figure 10 shows metabolite abundances stratified by chemotherapy/radiation status and surgery status.