| Part Ⅰ The Establishment of the the Discriminant Analysis Model and the Cross ValidationIntroduction and Objectives:Early detection of bone metastases is critical in the management of patients with high-risk PCa. The aim of this part is to search the correlative factors of bone metastasis in prostate cancer. Furthermore we established a novel screening model for BM in Chinese patients at the time when prostate cancer is diagnosed.Materials and Methods:The study included 488 patients who were diagnosed with PCa between 2009 and 2011 at a single center. All patients received bone scans using technetium 99m methylene diphosphonate at the initial staging. If the bone scan finding was equivocal, computed tomography or magnetic resonance imaging was performed to confirm the diagnosis. Age, prostate-specific antigen at diagnosis, clinical stage assigned according to the TNM 2002 staging system and biopsy Gleason score were collected in all patients. Multivariate logistic regression analysis was performed to identify statistically significant co-variates and then receiver operating characteristic curves were generated to identify optimal-cut-off values. Using these cut-off values, a formula was devised to calculate an index value for BM screening at diagnosis. The model was cross-validated using the leave-one-out method.Results:Of the 488 patients,65 patients (13.3%) had BM. cT4, Gleason Score≥4+3 and high PSA level are correlative factors of bone metastasis in prostate cancer. The area under the ROC curve was 0.87 (95% confidence interval= 0.83-0.94). The sensitivity of the cut-off point was 87.7%, and the specificity was 73.1%. Bone scan is needed for all cT4 PCa patients, however, it is advisable for cTl-T3 PCa patients who have a biopsy Gleason score≤3+4 and a PSA> 132.1 and for cTl-T3 patients having a Gleason score of≥4+3 and PSA>44.5. For cTl-T3 patients who have a Gleason Score≤3+4, if PSA≤132.1 ng/ml, the proportion of BM is 2%; if PSA>132.1 ng/ml, the proportion of BM is 25%. For cTl-T3 patients who have a Gleason Score≥3+4, if PSA<44.5 ng/ml, the proportion of BM is 4%; if PSA>44.5 ng/ml, the proportion of BM is 29%. For cT4 patients the proportion of BM is 73%. The sensitivity and specificity of cross validation were 86.2% and 71.9% respectively.Conclusion:cT4, Gleason Score>4+3 and high PSA level are correlative factors of bone metastasis in prostate cancer. The screening model established by this study can estimate theincidence rate of BM for different groups of patients. Score≤3+4, if PSA≤132.1 ng/ml, the proportion of BM is 2%; if PSA>132.1 ng/ml, the proportion of BM is 25%. For cTl-T3 patients who have a Gleason Score≥3+4, if PSA<44.5 ng/ml, the proportion of BM is 4%; if PSA>44.5 ng/ml, the proportion of BM is 29%. For cT4 patients the proportion of BM is 73%. The sensitivity and specificity of cross validation were 86.2% and 71.9% respectively.Conclusion:cT4, Gleason Score>4+3 and high PSA level are correlative factors of bone metastasis in prostate cancer. The screening model established by this study can estimate theincidence rate of BM for different groups of patients.Part Ⅱ The Optimization of the the Discriminant Analysis Model and the Exploration of Potential Risk FactorsIntroduction and Objectives:The aim of this study is to optimize the discriminant analysis model and to explore the potential risk factors.Materials and Methods:The study included 501 patients who were diagnosed with PCa between 2010 and 2013 at Zhongshan Hospital. All patients received bone scans using technetium 99m methylene diphosphonate at the initial staging. If the bone scan finding was equivocal, computed tomography or magnetic resonance imaging was performed to confirm the diagnosis. Age, prostate-specific antigen and alkaline phosphatase at diagnosis, clinical stage assigned according to the TNM 2002 staging system and biopsy Gleason score were collected in all patients. Multivariate logistic regression analysis was performed to identify statistically significant co-variates and then receiver operating characteristic curves were generated to identify optimal-cut-off values. Using these cut-off values, a formula was devised to calculate an index value for BM screening at diagnosis.Results:Of the 501 patients,84 patients (16.7%) had BM. The area under the ROC curve was 0.90 (95% confidence interval= 0.87-0.93). The sensitivity of the cut-off point was 94.1%, and the specificity was 58.3%. Baseline bone can be safety omitted for cTl-T3 PCa patients who have a PSA<39ng/ml and a ALP<88IU/.Conclusion:ALP can optimize the discriminant analysis model. The novel screening tool may help determine whether baseline bone scan can be omitted for newly diagnosed PCa patients. point was 94.1%, and the specificity was 58.3%. Baseline bone can be safety omitted for cTl-T3 PCa patients who have a PSA<39ng/ml and a ALP<88IU/.Conclusion:ALP can optimize the discriminant analysis model. The novel screening tool may help determine whether baseline bone scan can be omitted for newly diagnosed PCa patients.Part Ⅲ The External Validation of the Discriminant Analysis ModelIntroduction and Objectives:The aim of this study is to validate externally the discriminant anarysis model established in part Ⅱ to determine if the discriminant analysis model can be used in other databases.Materials and Methods:The discriminant analysis model was validated externally by the database of newly diagnosed PCa patients in Fudan University Shanghai Cancer Center, The database included 501 patients who were diagnosed with PCa between 2005 and 2011.The AUC of the ROC was compared with AUC of other screening tools.Results:External validation had an area under the ROC curve of 0.846(95% confidence interval= 0.805-0.887), which is better than the results of Fudan CART model, Briganti’s CART model, SRE model and ALP model.The sensitivity was 85.5%, and the specificity was 64.0%.Conclusion:The discriminant analysis model is reliable and can be used in other databases. |
PDF Full Text Request