| Objective:Using CLINPROT technology to screen serum proteome tumor markers and establish predictive models plus validation in colorectal cancer and CRC with liver metastasis patients.Methods:Drawing map of liquid serum protein chip:We chose 201 colorectal cancer cases from general surgery department in ZhongShan hospital affiliated by Fudan University between March,2004 and September, 2009. We established 3 disease groups, which are MLM (n=48), SLM (n=65), CRC (n=88). We also chose 45 normal cases as the control group. The serum samples were prepared from peripheral venous blood before operations. Special serum protein or peptide fingerprint was determined by MALDI-TOF-MS measurement after treating the samples onto Cu+ magnetic beads protein chip for each case. The obtained data were analyzed by CLINPROT software to screen serum proteome tumor markers in colorectal cancer and CRC with metastasis patients. We set up 4 subgroups:MLM, SLM, CRC, control, and performed comparisons; we also set up 2 subgroups: metachronous liver metastasis group, metachronous lung metastasis group and performed comparisons.Establishment of serum protein prediction model plus validation:We chose 28 MLM serum samples,39 SLM serum samples,49 CRC serum samples and 40 normal serum samples to establish predictive models plus validation in colorectal cancer and CRC with liver metastasis patients. In addition, we chose 5 MLM serum samples,26 SLM serum samples,39 CRC serum samples and 5 normal serum samples to perform blind validation. According to the first part mapping results, we set up 4 reliable liver metastases of colorectal cancer and colorectal cancer early diagnostic models, plus cross-validation and evaluation.4 predictive models were:CRC VS NORMAL; CRC VS SLM; MLM VS NORMAL; SLM VS NORMAL. 1. CRC VS MLM, we got 0 differentiate protein peaks, which showed no significant statistic difference (P>0.05)2. CRC VS NORMAL, we got 79 differentiate protein peaks, among which 15 protein peaks P＜0.000001; 44 protein peaks P＜0.01; 19 protein peaks P＜0.053. CRC VS SLM, we got 18 differentiate protein peaks, among which M/Z 1865.94 P＜0.000001; 8 protein peaks P＜0.01; 9 protein peaks P＜0.054. MLM VS NORMAL, we got 56 differentiate protein peaks, among which 5 protein peaks P＜0.000001; 21 protein peaks P＜0.01; 30 protein peaks P＜0.055. MLM VS SLM, we got 1 differentiate protein peaks, M/Z=4355.26 P=0.0366. SLM VS NORMAL, we got 88 differentiate protein peaks, among which 14 protein peaks P＜0.000001; 46 protein peaks P＜0.01; 28 protein peaks P＜0.057. metachronous liver metastasis group VS metachronous lung metastasis group, we got 0 differentiate protein peaks, which showed no significant statistic difference (P>0.05)1. CRC VS NORMAL:we got 70 differentiate protein peaks, CV(Cross Validation) was 92.25%, RC (Recognition Capability) was 96.71%.Sensitivity was 97.4%, specificity was 100%2. CRC VS SLM, we got 8 differentiate protein peaks, CV=69.06%,RC=76.30%. sensitivity was 90%, specificity was 56.40%3. MLM VS NORMAL, we got 54 differentiate protein peaks, CV=88.88%, RC=93.49%, sensitivity was 100%, specificity was 100%4. SLM VS NORMAL, we got 57 differentiate protein peaks, CV=87.08%, RC=90%, sensitivity was 100%, specificity was 60%1. Through 7 times comparisons, we got significant differentiate protein peaks 5 times, by combing with scatter distribution picture, we can establish 4 reliable CRC and liver metastases of CRC diagnostic models.2. The use of CLINPROT techniques to build models of serum protein patterns in predicting colorectal cancer and liver metastasis of colorectal cancer have a very high sensitivity and specificity, the forecast is expected to become very good diagnostic tool. 3. Through further isolation and identification, M/Z= 4644 with low expression or inactivation may be related with the occurrence and development of colorectal cancer; M/Z= 1865.94 protein is likely to predict synchronous liver metastatic colorectal cancer; M/Z= 4355.26 protein is likely to predict metachronous liver metastatic colorectal cancer. |
PDF Full Text Request