Multifactorial, site-specific recurrence model after radical cystectomy for urothelial carcinoma

BACKGROUND: A scoring algorithm of site-specific disease recurrence after cystectomy for urothelial carcinoma was designed. METHODS: Identified were 1388 patients who underwent radical cystectomy for nonmetastatic urothelial carcinoma between 1980 and 1998. Clinical, surgical, and pathologic features were evaluated for associations with 4 locations of site-specific disease recurrence: upper urinary tract, abdomen/pelvis, thoracic region, and bone. Recurrence-free survival rates were estimated using the Kaplan-Meier method. Cox proportional hazards models were fit to test associations with disease recurrence. RESULTS: A total of 493 (35.5%) patients experienced at least 1 recurrence. There were 67, 388, 143, and 145 patients with recurrences to the upper tract, abdomen/pelvis, thoracic region, and bone at a median of 3.1 years, 1.1 years, 1.3 years, and 1.0 years, respectively. Pathologic T4 stage (hazard ratio [HR], 2.84; P=.006), positive ureteral margins (HR, 5.71; P < .001), and multifocality (HR, 2.07; P=.009) were found to be independent predictors of upper tract recurrence. Pathologic T3 (HR, 2.30; P < .001) and T4 stage (HR, 3.55; P < .001), lymph node invasion (HR, 1.97; P < .001), extent of lymphadenectomy (pNx [HR, 1.66; P=.002] and <10 lymph nodes [HR, 1.52; P < .001]), multifocality (HR, 1.80; P < .001), and prostatic involvement (HR, 1.45; P=.019) were found to be independent predictors of abdominal/pelvic recurrence. Features independently associated with thoracic recurrence included pathologic T3 (HR, 2.61; P < .001) and T4 (HR, 3.39; P < .001), lymph node invasion (HR, 2.64; P < .001), extent of lymphadenectomy (pNx [HR, 1.89; P=.019] and <10 lymph nodes [HR, 1.58; P < .030]), and multifocality (HR, 1.79; P < .001). Pathologic T3 (HR, 3.45; P < .001) and T4 stage (HR, 3.87; P < .001), lymph node invasion (HR, 1.79; P=.006), occupational exposure to radiation (HR, 2.97; P=.003), and a positive urethral margin (HR, 2.28; P=.039) were found to be independent predictors of osseous recurrence. Macroscopic hematuria (HR, 0.52; P=.009) and obesity (HR, 0.59; P=.027) were found to be protective and negatively associated with upper tract and osseous recurrence, respectively. Scoring algorithms to predict the likelihood of disease recurrence to these sites were developed using regression coefficients from the multivariable models. CONCLUSIONS: Scoring algorithms based on independent predictors of site-specific recurrence were presented. These models may be used to tailor postoperative surveillance to the individual patient based upon clinicopathologic features at the time of cystectomy.