by Georg Johnen, Katarzyna Gawrych, Heike Bontrup, Beate Pesch, Dirk Taeger, Séverine Banek, Matthias Kluckert, Harald Wellhäußer, Friedhelm Eberle, Michael Nasterlack, Gabriele Leng, Arnulf Stenzl, Thomas Brüning, and the UroScreen Study Group
Urinary biomarkers have the potential to improve the early detection of bladder cancer. Most of the various known markers, however, have only been evaluated in studies with cross-sectional design. For proper validation a longitudinal design would be preferable. We used the prospective study UroScreen to evaluate survivin, a potential biomarker that has multiple functions in carcinogenesis. Methods/Results
Survivin was analyzed in 5,716 urine samples from 1,540 chemical workers previously exposed to aromatic amines. The workers participated in a surveillance program with yearly examinations between 2003 and 2010. RNA was extracted from urinary cells and survivin was determined by Real-Time PCR. During the study, 19 bladder tumors were detected. Multivariate generalized estimation equation (GEE) models showed that ß-actin, representing RNA yield and quality, had the strongest influence on survivin positivity. Inflammation, hematuria and smoking did not confound the results. Survivin had a sensitivity of 21.1% for all and 36.4% for high-grade tumors. Specificity was 97.5%, the positive predictive value (PPV) 9.5%, and the negative predictive value (NPV) 99.0%. Conclusions
In this prospective and so far largest study on survivin, the marker showed a good NPV and specificity but a low PPV and sensitivity. This was partly due to the low number of cases, which limits the validity of the results. Compliance, urine quality, problems with the assay, and mRNA stability influenced the performance of survivin. However, most issues could be addressed with a more reliable assay in the future. One important finding is that survivin was not influenced by confounders like inflammation and exhibited a relatively low number of false-positives. Therefore, despite the low sensitivity, survivin may still be considered as a component of a multimarker panel.