Fig. 1

Sera levels of several proteins in response and resistant to nCRT groups of rectal cancer. A-C PRM traces for the quantification of PZP, GELS and HEP2. D Scatter plots of sera PZP concentration obtained from pCR and non-pCR patients using the PRM. E Receiver operating characteristic (ROC) curve for nCRT response by the baseline PZP, GELS, HEP2 and CEA. ROC curves generated using the response information and levels of the baseline PZP, GELS, HEP2 and CEA are able to discriminate between pCR patients and non-pCR. PZP has the strongest predictive value (area under the curve [AUC] = 0.7993) to discriminate those patients. F ROC analysis for the training set using Random Forest. G ROC analysis for the testing set using Random Forest