Tumor markers in the diagnosis of primary bladder cancer. A systematic review

Purpose: We systematically reviewed the available evidence, and obtained and compared summary estimates of the sensitivity and specificity of cytology and the urine based markers bladder tumor antigen, BTA stat (Polymedco, Redmond, Washington), BTA TRAK (Polymedco), NMP22 (Matritech, Cambridge, Massachusetts), telomerase and fibrin degradation product in detecting primary bladder cancer. Materials and Methods: Studies on the diagnosis of primary bladder cancer published from 1990 through November 2001 in English and German were retrieved from MEDLINE and EMBASE data bases. In our research we included studies that evaluated I or more of the markers, used cystoscopy as the reference standard and allowed the construction of a 2x2 contingency table for a per patient analysis. The data plus items on study and clinical characteristics were extracted by 2 observers. Sensitivity and specificity for each marker were estimated using a bivariate random effect meta-analysis. A multivariable analysis was performed to explain study variation. Results: A total of 42 studies were included in our review. Only 2 studies were available on fibrin degradation product, hence a meta-analysis was not possible. Cytology had the best specificity at 94% (95% CI: 90% to 96%). This figure was significantly better than that of the other markers except for telomerase (specificity 86% [71% to 94%]). Telomerase had the best sensitivity (75% [71% to 79%]) but it was not significantly better than that of BTA stat (70% [66% to 74%]). Case control designs yielded lower values for sensitivity for the tumor markers cytology, bladder tumor antigen and BTA stat. Conclusions: Cytology has the best specificity and telomerase the best sensitivity. However, none of the markers studied here is sensitive enough to be recommended for daily routine.