We thank Dr. Sier and colleagues for their candid letter regarding our study.1 We chose not to control for kidney function; this decision was based on the assumption that for a marker to be clinically useful and to add predictive information to the information yielded by current tests, the marker should have a robust performance in all patients regardless of kidney function. We found that increased urine levels of urokinase (uPA) were associated with an increased risk of bladder carcinoma, whereas urine levels of the urokinase receptor (uPAR) were not. Urine levels of uPA have not been shown to fluctuate from sampling to sampling. Therefore, we concluded that urine levels of uPA represent a potentially valuable marker for bladder carcinoma detection. We found no association between urine levels of uPAR and the presence of bladder carcinoma or pathologic features. In our opinion, this finding, together with the dependence of urine levels of uPAR on sampling time (as Dr. Sier and colleagues appropriately have noted), supports the idea that urine levels of uPAR possess limited clinical utility in the diagnosis, staging, and management of bladder carcinoma. To address the specific issue raised by Dr. Sier and colleagues, we retrieved the majority of the samples used in the original study and repeated the analysis with adjustments for creatinine levels. Using a system manufactured by Olympus America (Melville, NY), we measured creatinine levels for 170 of the original 229 patients in the voided urine samples used for measurement of uPAR and uPA levels. The mean urine creatinine concentration was standard deviation (SD) 73.1 ± 46.4 mg/dL (median, 65.2 mg/dL; range, 9.8–236.1 mg/dL). All samples were run in duplicate, with the mean value used for data analysis. Differences between duplicate measurements were minimal, as indicated by the intraassay precision coefficient of variation, which was equal to SD 5.8 ± 8.7%. Levels of uPA and uPAR (both in ng/mL) were divided by urine levels of creatinine (in mg/dL) to adjust for the effect of kidney function. We found that both the uPA/creatinine ratio (P < 0.001) and the uPAR/creatinine ratio (P = 0.010) were greater in patients with bladder tumors compared with control patients. However, whereas there were significant differences between case patients and control patients in terms of the quartile distributions of the uPA/creatinine ratio (P = 0.002), there were no such differences with respect to the uPAR/creatinine ratio. In addition, uPA/creatinine ratios were higher among patients with abnormal urine cytology findings (P = 0.038), whereas uPAR/creatinine ratios were not associated with any clinicopathologic characteristics. The uPAR/creatinine ratio was significantly (but moderately) correlated with urine uPAR levels (correlation coefficient [CC], 0.496; P < 0.001), and the uPA/creatinine ratio was strongly correlated with urine uPA levels (CC, 0.904; P < 0.001). In univariate logistic regression analyses, uPA/creatinine ratio (P < 0.001), age (P < 0.001), and positive urinary cytology (P = 0.006), but not uPAR/creatinine ratio, were found to be associated with an increased risk of transitional cell carcinoma of the bladder. Using a multivariate logistic regression model, with adjustments made for the effects of urine uPAR levels and age, only increased uPA/creatinine ratio (P = 0.003) and positive cytology (P < 0.001) were found to be associated with the presence of bladder carcinoma. In conclusion, we confirmed that increased urine levels of uPA are associated with the presence of bladder carcinoma, regardless of kidney function, and that urine levels of uPAR are not associated with bladder carcinoma, even after adjustments for urine levels of creatinine are made. Shahrokh F. Shariat M.D.*, Seth P. Lerner M.D. , Roberto Casella M.D. , * Department of Urology University of Texas Southwestern Medical Center Dallas, Texas, Scott Department of Urology Baylor College of Medicine Houston, Texas, Department of Urology University Hospital Basel, Switzerland.
Read full abstract