A survival regression analysis of prognostic factors in colorectal cancer

Many prognostic factors of colorectal cancer are known but their actual clinical validity is still uncertain. The aim of the present study was to verify, on the basis of our experience, the prognostic validy of variables for survival by using survival regression analysis.

From January 1978 to December 1986 the prognostic factors for 192 patients were analysed. These patients had undergone surgical resection for colorectal cancer. The follow up was completed in every patient by the end of December 1992, with a median follow up of 10 years (range 6-14 years). The prognostic factors considered in the statistical analysis were age, sex, size of tumour, site, grade, direct spread, node involvement and stage (according to Astler-Coller and pTNM).

Of the prognostic factors, sex was the only one not to show any prognostic significance. In the survival regression analysis we have used an accelerated failure time model (equivalent to the Cox proportional hazard model); age, grade and stage were significant covariables.

Although clinical pathological staging (pTNM) appears as a pre-eminent prognostic factor, and as our analysis shows, it needs a further variable (grading), which has been shown to affect the prognosis in a significant way.