| Worldwide, Colorectal cancer is the third most frequently occurring cancer in both sexes; it ranks second in developed countries. In China, Colorectal cancer is the fourth leading cause of cancer mortality in big cities, the fifth in the countryside. It has been a major cancer and will be increasing in the near future. Many patients had evidence of locally advanced or metastatic Colorectal cancer at the time of initial presentation; only half of those who underwent apparently curative resection survived 5 years. The causative reason of death is associated directly with stage and therapeutic methods. The lack of good serum tumor markers causes great difficulty in determining its molecular stage preoperatively. Several common medical examination tools such as endoscopy, MRI, CT, etc. also cannot differentiate stages accurately. It is very important to oncologists to comprehend the stage before surgery in order to choose surgical schemes and formulate rational plans. Furthermore, presently the evaluation of curative effects, the estimation of prognosis and follow-up mainly depend on the before-mentioned, traditional methods too. Nevertheless, they are also not satisfying. It is estimated that half of patients with Colorectal cancer will develop liver metastases in the course of their disease. Unfortunately, only 10% to 20% of patients in this situation will benefit from surgical resection. Resection of resectable liver metastases from Colorectal cancer is the only way to cure. So, it is essential to screen resectable liver metastases timely.In some sense, these problems are due to the lack of good tumor biomarkers for Colorectal cancer in fact. Some tumor markers, such as the measurement of CEA and CA19-9 levels, have low sensitivities and/or specificities. Therefore, it is urgent to find sensitive, specific, and convenient new biomarkers.Protein but not nuclear acid is the material executant and embodiment of life. So, the studies on proteomics are preferred to approach to the pathogenesis of Colorectal cancer and screen sensitive and specific biomarkers. As a new proteomics approach,the ProteinChip based on surface enhanced laser desorption/ionization-time of flight-mass Spectrometry (SELDI-TOF-MS) could bind the proteins in the samples unselectively. It combines ProteinChip array with time-of-flight mass Spectrometry and offers the advantages of speed, simplicity, sensitivity and suitability for a comparative study. It can directly obtain high-throughput protein profilings from clinical samples with high sensitivity and this is the main advantage of this technology. To look for such "fingerprints" of cancer, it will require not only high-throughput genomic or proteomic profiling, but also sophisticated bioinformatics tools for complex data analysis and pattern recognition. In proteomics, the technology of bioinformatics carved out a new way to effectively seek tumor markers.Three parts compose this study. Using SELDI-TOF-MS technology and bioinformatics tools, firstly, to detect the serum proteomic patterns in Colorectal cancer patients, and to evaluate the significance of the proteomic patterns in the tumour staging preoperatively. Secondly, to seek the cancer progression tumor markers in patients' serum postoperatively and to instruct tumor therapy and evaluate prognosis. Thirdly, to seek the liver metastases biomarker patterns in Colorectal cancer patients postoperatively. The purposes of this study are to screen novel tumor biomarkers of Colorectal cancer that are fit for staging and monitoring of recurrence and metastasis, especially for liver metastases.Part 1: The Application of Serum Protein Fingerprints in the staging of Colorectal CancerIn this study, we screened the serum proteomic patterns by using SELDI-TOF-MS technology and CM10 ProteinChip in Colorectal cancer (CRC) patients and evaluated the significance of the proteomic patterns in the tumour staging of Colorectal cancer. A total of 76 serum samples were obtained from CRC patients at different clinical stages, including stage â… (n=10), stage â…¡ (n=19), stage â…¢ (n=16) and stage â…£ (n=31). The biomarker wizard of ProteinChip Software 3.2 was used to compare the data of different group by Mann-Whitney U test and discrepant mass peaks werefound. The samples of different stage groups models were developed and validated by discriminant analysis, support vector machines (SVM) and time-sequence analysis. Training was conducted to convergence on the training data and minimized the errors. These statistical analysis tools were performed by MATLAB-NNTools software (The Math Works Inc., USA). Training was conducted to convergence on the training data and minimized the errors.The discriminant analysis and support vecter machines models introduced random perturbations in multiple runs to test the consistency of the top 10 ranked peaks, measured by the P value of m/z peaks of computed ranks from multiple runs. Then stage models were built by using the selected peaks. The models established based on these selected biomarkers should be further validated independently. Here, Leave-one-out Cross-validation approach was applied to estimate the accuracy of the classifier to determine the misclassification rate. Time-sequence analysis was used to distinguish different stage groups.The results of our study showed that six protein peaks of 2760, 2965, 2048, 4796, 4140 and 37762 m/z were top-scored and finally selected as the potential biomarkers for the distinguishing of local CRC patients (stage I and stage â…¡) from regional CRC patients (stage â…¢) with the specifity of 86.21%, sensitivity of 87.50% and accuracy of 86.67%. Three protein peaks (6885, 2058 and 8568 m/z) could be used to distinguish locoregional CRC patients (stage â… , stage â…¡ and stage â…¢) from systematic CRC patients (stage â…£) with the specifity of 71.11%, sensitivity of 80.65% and accuracy of 75.00%. And we also distinguished stage â… from stage â…¡ with an accuracy of 86.21% (25/29), stage â… from stage â…¢ with an accuracy of 84.62%(22/26), stage â…¡ from stage â…¢ with an accuracy of 85.71%(30/35), stage â…¡ from stage â…£ with an accuracy of 80.00% (40/50), stage â…¢ from stage â…£ with an accuracy of 78.72%(37/47). In addition, patients with liver metastasis of stage â…£ were differentiated from those of adjacent local celiac metastasis and systemic distant metastasis. Different stage groups could be distinguished by the two-dimensional scattered spots figure obviously.Part 2: The application of Serum Protein Fingerprints in the Postoperative Follow-up of Colorectal CancerSeventy-two postoperative CRC patients were followed-up, including 42 patients diagnosed as postoperative recurrences or metastases and 30 patients diagnosed as no postoperative recurrences or metastases. Followed-up time were from April 2002 to April 2005.The median time was 16 months. The serum protein fingerprints of the 72 samples were detected. The data were pretreated and the discrepant proteins were screened to identify cancer progressive biomarkers by the same methods in part one. The protein fingerprints of the 30 patients survived free from cancer after their operations and the 42 patients with postoperative recurrences and metastases were compared by Biomarker Wizard 3.2.0. software. Eleven discrepant proteins were screened in the comparison (P<0.05). The expression of 44888, 5908, 28092, 5341, 2957and 8566m/z of the 30 patients who survived free from cancer after their operations were higher than that of the patients with cancer progression after their operations (P<0.05). Whereas, five proteins (2107, 6114, 2131, 2162 and 2389 m/z) were up regulated when cancer progession than those without cancer progession (P<0.05). Using SPSS 12.0 software, we finally selected 2 from 11 proteins whose AUC values were more than 0.7.Part 3: The application of Serum Protein Fingerprints in the Postoperative liver metastases of Colorectal CancerThe serum protein fingerprints of the 72 samples above were analysed furtherly. These 72 postoperative patients were grouped by 19 had liver metastases as the only site of metastases (Group A), 7 had liver metastases and concomitant other sites metastases or recurrence (Group B), 23 had any other metastases or recurrence sites with or without liver metastases (Group C) and 30 patients survived free from cancer (Group D).The SVM approach and Leave-one-out Cross-validation were used. The results showed that 4 protein peaks of 6199, 5482, 9295and 2872 m/z were finally selected as the potential biomarkers to distinguish Group A from Group B with the accuracy of92.31 % (24/26). Four protein peaks of 6199, 3419, 2399 and 2127 m/z were finally selected as the potential biomarkers to distinguish Group A from Group C with the accuracy of 83.33% (35/42). Other four protein peaks of 2208, 2299, 2355 and 2758 m/z were finally selected as the potential biomarkers to distinguish Group A from Group D with the accuracy of 87.76% (43/49).The expression of 6199 m/z of the patients with liver metastases were higher than that of the patients with other sites metastases or recurrence (P<0.05). The 2208, 2299, 2355 and 2758 m/z were up regulated in patients with liver metastases than those without cancer progession (P<0.05).ConclusionsTo discover new biomarkers and establish patterns for the CRC cancer staging preoperatively.1. Six protein peaks of 2760, 2965, 2048, 4796, 4140 and 37762 m/z were top-scored and finally selected as the potential biomarkers for the distinguishing of local CRC patients (stage â… and stage â…¡) from regional CRC patients (stage â…¢) with the specifity of 86.21%, sensitivity of 87.50% and accuracy of 86.67%.2. Three protein peaks (6885, 2058 and 8568 m/z) could be used to distinguish locoregional CRC patients (stage â… , stage â…¡ and stage â… ) from systematic CRC patients (stage â…£) with an accuracy of 75.00%.3. Also distinguished stage â… from stage â…¡ with an accuracy of 86.21% (25/29), stage I from stage III with an accuracy of 84.62% (22/26), stage â…¡ from stage III with an accuracy of 85.71% (30/35), stage â…¡ from stage â…£ with an accuracy of 80.00% (40/50), stage â…¢ from stage â…£ with an accuracy of 78.72%(37/47).4. Different stage groups could be distinguished by the two-dimensional scattered spots figure obviously.To discover new biomarkers for the CRC cancer screening postoperatively.1. Eleven discrepant proteins were screened in the comparison (P<0.05).2. The expression of 44888, 5908, 28092, 5341, 2957and 8566m/z of the 30 patients who survived free from cancer after their operations were higher than that of the patients with cancer progression after their operations (P<0.05).3. Five proteins (2107, 6114, 2131, 2162 and 2389 m/z) were up regulatedwhen cancer progession than those without cancer progession (P<0.05).4. Two from eleven proteins were selected whose AUC values were more than 0.7.To discover new biomarkers for the CRC cancer liver metastases postoperative.1. Four protein peaks of 6199, 5482, 9295and 2872 m/z were finally selected as the potential biomarkers to distinguish liver metastases from liver and concomitant other sites metastases or recurrence with the accuracy of 92.31% (24/26).2. Four protein peaks of 6199, 3419, 2399 and 2127 m/z were finally selected as the potential biomarkers to distinguish liver metastases from any other metastases or recurrence sites with the accuracy of 83.33% (35/42).3. Other four protein peaks of 2208, 2299, 2355 and 2758 m/z were finally selected as the potential biomarkers to distinguish liver metastases from patients who survived free from cancer with the accuracy of 87.76% (43/49).4. The expression of 6199 m/z of the patients with liver metastases were higher than that of the patients with other sites' metastases or recurrence (P<0.05).5. The 2208, 2299, 2355 and 2758 m/z were up regulated in patients with liver metastases than those without cancer progession (P<0.05).
|
PDF Full Text Request