Assessing real-world recurrence in high-risk (HR) non-muscle-invasive bladder cancer (NMIBC) treated with bacillus Calmette-Guérin (BCG) in the United States through a recurrence algorithm: A SEER-Medicare study.

Abstract

4601 Background: Assessing bladder recurrences in real-world setting can be challenging due to the lack of readily available data on recurrence in large real-world databases (e.g., SEER-Medicare), which limits the ability to evaluate the long-term real-world outcomes. To address this, this study developed an algorithm using linked SEER-Medicare data (2007–2020) to identify and classify recurrence events. Methods: A retrospective cohort study in patients with HR-NMIBC treated with BCG was conducted, with BCG initiation as the index date. A recurrence algorithm was developed to identify and classify recurrence events including NMIBC recurrence, muscle-invasive bladder cancer (MIBC) progression, and distant metastasis (DM). NMIBC recurrence and MIBC progression were identified based on repeat TURBT procedures occurring ≥30 days after the last BCG treatment, supplemented by cancer diagnoses in the urethra or upper tract. Subsequent treatments were used to further classify them as NMIBC recurrence (intravesical BCG after a ≥6-month gap from the last BCG, or intravesical chemotherapy) and MIBC progression (systemic therapy, radiotherapy, or cystectomy). DM was identified by urothelial cancer diagnoses outside the urinary bladder, urethra, or upper tract. Cumulative incidence rates from the index date for each recurrence type were calculated accounting for competing events, such as death and more severe types of recurrences (e.g., MIBC or DM for NMIBC recurrence). Results: A total of 5,490 patients (median follow-up: 2.9 years) were included. Median age at index diagnosis was 76.5 years. NMIBC recurrence was the most common type of recurrences, followed by MIBC progression and DM. The cumulative incidence rates were 15.9% for NMIBC recurrence, 4.1% for MIBC progression, and 3.7% for DM at 1 year, reaching 33.6%, 15.9%, and 16.7% at 10 years, respectively (Table). Conclusions: Using linked SEER-Medicare data, an algorithm was developed to identify and classify disease recurrences, facilitating long-term outcomes analyses in a large, real-world cohort. Notably, most recurrences occurred within the first 5 years from BCG initiation, after which rates plateaued. These findings underscore the importance of routine imaging for early detection and timely intervention and the need for better treatments to reduce recurrence and progression burden in this population. Cumulative incidence rates by type of recurrences (N = 5,490) 1 . Recurrence type Number of events during the follow-up period 1 year 3 years 5 years 7 years 10 years NMIBC recurrence 1528 15.9% 28.7% 31.6% 32.6% 33.6% MIBC progression 552 4.1% 9.3% 12.1% 14.0% 15.9% DM 528 3.7% 8.1% 11.5% 13.5% 16.7% 1 Patients were censored at the end of continuous eligibility or the end of data availability, whichever occurred earlier.