Risk estimation of multiple recurrence and progression of non muscle invasive bladder carcinoma using new mathematical models.

Abstract Objective To apply new mathematical models according to Non-Muscle Invasive Bladder Carcinoma (NMIBC) biological characteristics and enabling an accurate risk estimation of multiple recurrences and tumor progression. The classical Cox model is not valid for the assessment of this kind of events because the time between recurrences in the same patient may be strongly correlated. These new models for risk estimation of recurrence/progression lead to individualized monitoring and treatment plan. Materials and methods 960 patients with primary NMIBC were enrolled. The median follow-up was 48.1 (3–160) months. Results obtained were validated in 240 patients from other center. Transurethral resection of the bladder (TURB) and random bladder biopsy were performed. Subsequently, adjuvant localized chemotherapy was performed. The variables analyzed were: number and tumor size, age, chemotherapy and histopathology. The endpoints were time to recurrence and time to progression. Cox model and its extensions were used as joint frailty model for multiple recurrence and progression. Model accuracy was calculated using Harrell's concordance index (c-index). Results 468 (48.8%) patients developed at least one tumor recurrence and tumor progression was reported in 52 (5.4%) patients. Variables for multiple-recurrence risk are: age, grade, number, size, treatment and the number of prior recurrences. All these together with age, stage and grade are the variables for progression risk. Concordance index was 0.64 and 0.85 for multiple recurrence and progression respectively. Conclusion The high concordance reported besides to the validation process in external source allows accurate multi-recurrence/progression risk estimation. As a consequence, it is possible to schedule a follow-up and treatment individualized plan in new and recurrent NMCB cases.