Prediction of multifocal lesions in patients with upper tract urothelial carcinoma

To identify predictive factors for multifocal lesions in patients with upper tract urothelial carcinoma (UTUC).

Between June 1996 and September 2013, a cohort of 147 patients underwent radical nephroureterectomy at our institution. Univariate and multivariate logistic regression analyses were performed to identify clinical covariates significantly associated with the multifocal UTUC lesions and establish a model to predict its occurrence.

In the univariate analyses, estimated glomerular filtration rate, hydronephrosis, history of bladder cancer, hemoglobin level, and Brinkman index were significantly associated with multifocal lesions. In the multivariate analysis, hydronephrosis (P = .010), history of bladder cancer (P = .017), Brinkman index >100 (P = .014), and hemoglobin level <11 g/dL (P = .031) were independent predictive factors for multifocal lesions. Using these independent factors, we calculated the predictive probability of multifocal lesions and predicted their occurrence. The predictive probability of multifocal lesions strongly correlated with their observed rate (regression coefficient = 0.9999).

An improved understanding of UTUC has led to various clinical guidelines being revised to recommend less invasive interventions in appropriate cases. Hydronephrosis, history of bladder cancer, Brinkman index, and hemoglobin level are independent predictive factors for multifocal lesions of UTUC. Although this study has limitations and additional studies are required to validate our findings, this information may aid in appropriate patient selection for partial ureterectomy and may be effectively incorporated into clinical practice.