| Objective The purpose of this study is to analyze the different serum proteomics between colorectal cancer patients and controls,preoperative colorectal cancer patients and postoperative colorectal cancer patients,colorectal cancer patients with and without metastasis using SELDI-TOF-MS, in order to develop a proteomic pattern for distinguishing individuals with colorectal cancer from healthy controls and monitoring metastasis.Methods The serum of colorectal cancer patients,benign colorectal disease and healthy volunteers was collected and divided into two sets. The training set consisting of 63 patients with colorectal cancer,20 patients with benign colorectal diseases and 26 healthy volunteers was used to develop a proteomic model that discriminated colorectal cancer effectively. The sensitivity and specificity of this model was validated by an independent test set. To explore serum proteins changed after operation, the protein profiles of 31 postoperative patients were compared with those of preoperative patients. We also analyzed protein profiles of patients with and without metastasis to monitor metastasis.Results1. ReproducibilityThe reproducibility of SELDI mass spectra was successfully testified using the quality control (QC) samples. The intra- and inter-assay coefficients of variance for peak location were 0.04% and 0.05%, and the intra- and inter-assay coefficients of variance for normalized intensity (peak height or relative concentration) were respectively 11% and 14%. There was little variation with day-to-day sampling and instrumentation. The acceptable intra- and inter-assay variations of this method have allowed us to obtain a reliable result in this study.2. Cancer-specific biomarkers detection and selectivityA total of 127 peaks were identified in the m/z region of 2000～30000 from SELDI spectra of training set. Using Biomarker Wizard software, we compared the spectra generated from cancer group with corresponding spectra generated from control group. This comparison yielded 26 differential peaks. Among these, 4 peaks were chose to form a model that could discriminate colorectal cancer patients from control group effectively. The 4 peaks corresponded to m/z ratios of 3191.5,3262.9,3396.3 and 5334.4. All the 4 peaks were up-regulated in the group of patients with colorectal cancer (P < 0.01). The sensitivity and specificity of this model was respectively 90.3% and 95.7%. A blind test set consisted of 48 patients with colorectal cancer, 18 patients with benign colorectal neoplasia and 14 healthy volunteers. In our study, correctly classification was achieved in 30 of 32 controls and 42 of 48 cancer patients, including 8 of 10 Duke'A patients.3.Different preoperative and postoperative markers in colorectal cancerWe compared the preoperative protein profiles with the postoperative (day 14) profiles for the 31 colorectal cancer patients. Two peaks (m/z: 2753.8 and 4172.4) were detected which were down-regulated in 27 of 31 (87.5%)patients compared to these in preoperative samples. In an independent test set, the two peaks were also validated down-regulated in 13 of 16 (81.3%) postoperative samples.4.Differential markers for primary colorectal cancer and metastatic colorectal cancerThe cancer patients of the training set were divided into two groups (30 patients with metastasis and 33 patients without metastasis) according to after surgical examination. Two proteins (m/z: 9184.4 and 9340.9) were found that can discriminate the two groups. The two proteins were observed in all the Duke'A and in 13 of 15 Duke'B patients, absent in 11 of 14 Duke'C and all Duke'D patients from the test set.5. Result of CA199, CA242 and CEA levels assay CA199, CA242 and CEA levels were available for all the cases in training set and test set. We found that combination of these three markers had the sensitivity of 62.4% and specificity of 86.2% for distinguishing colorectal cancer from controls. Obviously, the proteomic model generated from our study had higher sensitivity than the combination of CA199, CA242 and CEA for diagnosing colorectal cancer (P < 0.005) though the specificity had no statistic difference.Conclusions1. It is reliable to obtain proteomic profiles in serum for colorectal cancer using SELDI-TOF-MS with well reproducibility.2. The diagnostic model composed of 4 differentially expressed proteins(m/z of 3191.5,3262.9,3396.3 and 5334.4) have the ability to distinguish colorectal cancer from healthy control. It performed significantly better than the combined use of CEA,CA199and CA242 in sensitivity.3. Two peaks (m/z: 2753.8 and 4172.4) were detected down-regulated in postoperative samples than preoperative samples. We hypothesized that these two biomarkers may be oncogene proteins and provide a new insight into therapeutic strategies and molecular mechanism behind the process of tumorigenesis.4. Two peaks (m/z: 9184.4 and 9340.9) were found that can discriminate the primary colorectal cancer group with metastatic colorectal cancer group. We concluded that these two biomarkers may be metastasis related proteins and can be used to monitor metastasis at the early stage.
|
PDF Full Text Request