COX Proportional Hazards Model Analysis Of Prognostic Factors Impacting On Survival Of Prostate Cancer

Objective The aim of the study is to explore prognostic factors and establish COX proportional hazards model of prostate cancer,providing proof for predicting prostate cancer prognosis and clinical treatment.Methods A data-base of 157 patients with prostate cancer diagnosed by prostate biopsy guided by TRUS between 1995 and 2003 were analyzed.Overall survival rate,disease specific survival rate and progress-free survival rate for patients were analysed.Thirty factors possiblly impacting on prognosis were selected and statisticly analyzed.Survival curve was drew by Kaplan-Meier method. Independent prognosis factors were screened and predictive equation by COX proportional hazards model.PI(Prognostic index) threshold was identified and verified.Results(1) Overall survival rates at 1-,3-,5-,8- and 10- year were 91%, 77%,72%,62%,58%,corresponding disease specific survival rats and progress free survival rates of 94%,85%,82%,76%,71%,and 91%,76%,70%,63%, 59%,respectively.(2) The independent factors impacting on progress-free survival included seminal vesicle involvement,the percentage of prostate biopsy cores with cancer,Gleason score,radical prostatectomy and the way of a tumor fund.The prognosis index equation was created as PI=0.216X₇+0.590X₁₂+0.356X₁₃-0.655X₂₄-0.989X₂₅. Chi-square test showed X²=52.843 and p=0.000,which illustrated the equation reasonable.PI threshold was-1.0 identified according to the equation.There were basically agreement between predictive results and the actual results.The sensitivity,specificity and accuracy were 91.07%,90.70%and 90.34%.The high risk group and low risk group were classified by PI threshold.(3) The independent factors impacting on disease specific survival included the percentage of prostate biopsy cores with cancer,Gleason score,radical prostatectomy,the way of a tumor fund and patient origin.The prognosis index equation was created as PI=0.918X₁₀+0.635X₁₂+0.438X_13-0.647X₂₄-0.776X₂₅-0.331X₃₀,Chi-square test showed X²=97.302 and p=0.000,which illustrated the equation reasonable.PI threshold was -0.4 identified according to the equation.There were basically agreement between predictive results and the actual results. The sensitivity,specificity and accuracy were 89.17%,72.00%and 86.21%.The high risk group and low risk group were classified by PI threshold.(4) The independent factors impacting on overall survival included metastases,the percentage of prostate biopsy cores with cancer,Gleason score,T stage and radical prostatectomy.The prognosis index equation was created as PI=0.403X₈+0.694X₁₀+0.768X₁₂+0.468X₁₃-1.124X₂₅,Chi-square test showed X²=91.24 and p=0.000,which illustrated the equation reasonable. PI threshold was 2.4 identified according to the equation.There were basically agreement between predictive results and the actual results.The sensitivity,specificity and accuracy were 86.36%,80.00%and 84.83%.The high risk group and low risk group were classified by PI threshold.Conclusions(1)The 157 patients prognosis was better,which might attributed to prostate biopsy guided by TRUS.(2)Radical prostatectomy,the way of a tumor fund and patient origin were protective factors of prostate cancer prognosis,while seminal vesicle involvement,percentage of prostate biopsy cores with cancer, Gleason score,metastases and T stage were danger factors of prostate cancer prognosis.(3) The survival equation and PI can accurately predict patient prognosis,thus providing a useful tool for clinical treatment.