Optimization Of Diagnostic Model And Identification Of Biomarkers For Colorectal Cancer And Liver Metastases

Objective:Accurate and reliable biomarkers for the early diagnosis of colorectal cancer (CRC) remains an urgent need, so we used magnetic bead fractionation coupled with MALDI-TOF MS to screen serum samples to discover biomarkers.Methods:We used46CRC serum samples and35healthy controls to build the diagnostic model. The remaining45CRC serum samples,42controls and46from other cancer were tested for validation. To evaluate the robustness of the model, we further tested the peptide signatures on an external independent set of269serum samples from CRC patients and110from healthy controls. And the differential peptides were identified by LC-MS/MS.Results:The Fisher-based diagnostic model with5proteomic features (m/z1778.97,1866.16,1934.65,2022.46and4588.53) was generated with best performance in the training and test sets. The Fisher-based model could discriminate CRC patients from healthy controls with100%sensitivity and100%specificity in the training set, and95.6%sensitivity and83.3%specificity in the test set. And the model only classified54.3%of other cancer serum samples as positive. Finally we obtained94.4%(254/269) sensitivity,75.5%(83/110) specificity and88.9%(337/379) accuracy in the external independent set. Further, the first four proteomic peaks were identified as fragments of complement C3f with MS/MS.Conclusion:The Fisher-based model with above five proteomic peaks could effectively differentiate CRC patients from healthy controls and other cancer patients, and the model may be specific for CRC. Decrease of serum complement C3f in colorectal patients may be relevant to the occurrence of CRC. BackgroundThere are currently no accurate predictive markers of metachronous liver metastasis (MLM) from colorectal cancer.MethodsMagnetic bead-based fractionation coupled with mass spectrometry analysis was used to compare serum samples from64patients with MLM and64without recurrence or metastasis for at least3years after radical colorectal surgery (NM). A total of40MLM and40NM serum samples were randomly selected to build a decision tree, and the remainder were tested as blinded samples. Selected peptides were identified.ResultsThe patients in the two groups were matched for gender, age, tumor location, TNM staging, and histological differentiation grade. Preoperative serum Carcinoembryonic Antigen (CEA) retained no independent power to predict MLM. The decision tree model with8proteomic features (m/z3315,6637,1207,1466,4167,4210,2660, and4186) correctly classified33(82.5%) of40NM sera and32(80%) of40MLM sera in the training set, and19(79.2%) of24NM sera and17(70.8%) of24MLM sera in the test set. The peptides were identified as fragments of alpha-fetoprotein, complement C4-A, fibrinogen alpha, eukaryotic peptide chain release factor GTP-binding subunit ERF3B, and angiotensinogen.ConclusionsIn patients matched for gender, age, tumor location, TNM staging, and histological differentiation grade, preoperative CEA retained no independent power to predict MLM. The decision tree model of8proteomic features demonstrated promising value for predicting MLM in patients who underwent radical resection of colorectal cancer. Objective:Colorectal cancer could be cured by surgery at early stage. However, most patients were with advanced disease at diagnosis, and had poor prognosis. So it is urgent to diagnose colorectal cancer at early stage. The study was to identify biomarker for early diagnosis of colorectal cancer with tissue proteomics technology.Methods:We used2D SDS-PAGE and MS-based proteomics approach to study the differentially expressed proteins in tumor, adjacent non-tumor and normal tissue samples from5pathologically confirmed colorectal patients. Then GO analysis and pathway analysis of differentially expressed proteins were conducted with bioinformatics analysis. Significantly altered proteins were confirmed using Western blot analyses in18colorectal patients.Results:31significantly and consistently altered proteins were identified from50differentially expressed protein spots. GO analysis revealed that a high proportion of cytoplasmic and mitochondrial proteins in cellular components, a high proportion of protein binding, oxidoreductase activity, and thioredoxin peroxidase activity in molecular function, and a high proportion of oxidation reduction, cell redox homeostasis, anti-apoptosis, response to glucocorticoid stimulus in biological process. I-FABP was one of the most significantly altered proteins and its low expression in colorectal cancer was confirmed using Western blot analyses. HSP27was one of the most significantly altered proteins and its overexpression in colorectal cancer was confirmed using Western blot analyses.Conclusion:1.31differentially expressed proteins from colorectal cancer were identified with2D SDS-PAGE and MS-based proteomics approach2. Low expression of I-FABP and overexpression of HSP27were confirmed using Western blot analyses, and these data suggested that they may be potential biomarkers for early diagnosis of colorectal cancer.