Overexpression of CD44: a useful independent predictor of prognosis in patients with colorectal carcinomas.

Abstract

The goal was to investigate the potential correlation between overexpression of CD44, high microvessel count (MVC), and p21ras with length of relapse-free and overall survival in patients with colorectal adenocarcinomas. CD44, factor VIII-related antigen (FVIII-RA), and p21ras were localized immunohistochemically in patients with colorectal adenomatous polyps (n = 8) and adenocarcinomas (n = 98). The correlation between the expression of CD44, MVC in the areas with highest density, and p21ras with relapse-free and overall survival time was investigated. Data were analyzed statistically using univariate and multivariate systems. In patients with adenomatous polyps, the positivity of CD44, FVIII-RA, and p21ras was 75%, 62%, and 88%, respectively. In patients with colorectal carcinomas the positivity of CD44 was 55%, and for p21ras it was 52%. The median of FVIII-RA was 4 MVC (range, 0.0 to 32.33). MVC was greater than 4 in 53% of the patients with colorectal carcinomas. In univariate analysis, a significantly longer relapse-free time (CD44: P = .0004; FVIII-RA: P = .0006) and overall survival time (CD44: P = .0001; FVIII-RA: P = .001) were observed for patients with CD44-negative tumors and MVC below 4 as compared to those with CD44-positive tumors and MVC greater than 4. Similar observations were noted in patients with Dukes B and C disease and the rectum as the site of tumor. In multivariate analysis, only CD44 correlated significantly with both relapse-free (P = .0003) and overall survival (P = .00001). Univariate analysis showed CD44 and MVC to be independent predictors of prognosis in colorectal carcinomas. Multivariate analysis showed that CD44 positivity was the most important indicator of an unfavorable prognosis for relapse-free and overall survival in patients with colorectal cancer. Thus, it can be deduced that whether CD44 is positive or negative in patients with colorectal cancer may have prognostic importance and in the future may be used as a factor in the pathologic evaluation of tumor specimens. This hypothesis needs to be tested prospectively in a larger number of patients.