Multivariate Analysis Of Prognostic Factors Of Colorectal Cancer With COX Proportional Hazard Model

PurposeColorectal cancer occurs frequently worldwide, being the second most common cause of death from malignancy, both in men and women, in western countries. And its incidence rate was increasing steadily in recent. It's important to identify those prognostic factors, which is a primary basis for therapeutic protocol selection and predicting the outcome of patients with colorectal cancer. Using univariate and multivariate analyses to identify the prognostic value of the clinic-pathologic factors of colorectal cancer in our study, we expect to find some undocumented prognostic factors and create the individual risk assessment model for predicting the prognosis of the patients with colorectal cancer.Material and methodFrom 1980 to 1995, 1334 cases of colon and rectal cancer were operated and confirmed by pathology at the second affiliated hospital of Zhejiang university, and 1279 cases were collected from above patients in our study. The univariate and multivariate analyses were applied to evaluate the prognostic factors. The end point of the analysis was overall survival rate. The questionnaires were input to FOXBASE database after it was encoded, and thenread by SPSS 10.0 statistic software. On the univariate analysis, the cumulative survival rates were calculated with the Kaplan-Meier method; The Log-Rank test was applied for comparison. Potential prognostic factors were analyzed on a multivariate analysis using a forward step-wise non-conditional Cox proportional hazards model. All tests were performed at the 0.05 level of significances. Statistical analysis was completed using the SPSS 10.0 statistic software.ResultThe median follow-up time of the patients was 3.75 years (0~20.17ys). 692(54.11%) cases died of this malignancy, and 587(45.89%) censored. The distribution of the age was 53.66?4.62ys, 6.33% of the patients were younger than 30y, and 21.42% older than 65y. The overall survival rate for 3,5,10 years was 0.60+0.01,0.52 + 0.01,0.42+0.02 respectively; While the survival rate for 3,5,10 years after radical resection was 0.71?.02,0.63?.02,0.51 ?.02 respectively. From the univariate analysis, tumor site, obstruction, penetration, family history of colorectal cancer, radical resection, combined resection of other organ, chemotherapy during operation, tumor size, differentiation, histological type, intestinal wall involvement, age, adjacent organ involvement, distant metastasis, lymph node metastasis, total number of lymph nodes, perioperative blood transfusion, blood CEA level, blood albumin were found to be significantly associated with the cumulative survival rate. From the multivariate analysis, only lymph node involvement, distant metastasis, total number of lymph nodes and radical resection remained independently associated with cumulative survival rate.5ConclusionFrom our study, we concluded that:1. Lymph node metastasis, radical resection, distant metastasis, and total number of lymph nodes are the most important independent prognostic factors predicting the outcome of patients with colorectal cancer.2. When tumor infiltrates mucosa, submucosa, superficial muscularis or profound muscularis, the cumulative survival rates show no difference statistically; When tumor infiltrates the serosa, the cumulative survival rates decrease significantly; When tumor infiltrates through the serosa, the cumulative survival rates decrease significantly.3. Patients with lymph node metastasis have poorer prognosis; Patients with metastatic lymph nodes more than 3 have poorer prognosis, compared to those with 1~3 metastatic lymph nodes.4Patients with total lymph nodes more than 10, or with distant metastasis (found during operation) have poorer prognosis.