AUA 2018: Frequency of Low Grade T1 Bladder Cancer Has Decreased but Continues to Vary by Institution

San Francisco, CA ( Non-muscle invasive bladder cancer (NMIBC) encompasses approximately 70% of new bladder cancer diagnoses. The 2004 WHO classification system classifies tumors into low and high grade disease on the basis of microscopic characteristics. In contrast, the TNM staging classifies NMIBC by depth of invasion. In general, these two correlate to some degree in that most low grade (LG) disease is Ta (involving urothelium but not into lamina propria) disease and HG disease can be either Ta or T1 (into lamina propria). LG T1 disease is very uncommon, and to many urologists, is a flag for requiring pathology re-review. True LG T1 disease is quite uncommon and a management conundrum.

The authors of this abstract utilize the SEER and NCDB databases to evaluate the incidence of LG T1 disease over time. However, the inherent nature of this database ensures that there are numerous quality control issues with the abstract:

1. There is no pathology review for these diagnoses – many are in the community, and if re-reviewed, may be changed.
2. Despite focusing on patients after 2004 to ensure WHO 2004 classification, in real-life, that may not be the case.
3. SEER only captures the initial diagnosis – not any subsequent diagnoses.

Despite this, they assessed all patients diagnosed with T1 NMIBC from SEER (2004-2014) and NCDB (2004-2015). They trended the proportion of patients with LG disease over time. They also tried to identify variables that predicted LG T1 disease.

SEER analysis:

34,735 patients were diagnosed with T1 bladder cancer.  The percentage of LG cancers decreased from 33.4% in 2004 to 15.6% in 2014 (p < 0.001). Patients with HG T1 had worse 10-year CSS and OS, 70.9% and 42.6%, compared to LG, 83.1% and 51.5%, respectively (p < 0.001)

NCDB analysis: 

58,092 patients with T1 NMIBC, with 26.5% LG.  The percentage of LG decreased over time from 38.9% to 18.5% (OR 0.38 [95% CI, 0.35-0.41, p < 0.001] for 2013 vs 2004). 

Predictors of LG T1 diagnosis:

On logistic regression, each successive year, the type of institution (academic or comprehensive compared to the community) and distance from treatment facility greater than 60 miles were associated with lower rates of LG. Within NCDB institutions, the proportion of LG T1 ranged from 0% to 92%.  Among LG T1 patients, there was no difference in survival when stratified by year of diagnosis.

Ultimately, I agree with their take-home point that the institution is the biggest predictor of LG T1 diagnosis – it is less likely to be identified in an academic center or urban areas. With pathology review and dedicated GU pathologists, these are likely being appropriately reclassified as LG Ta disease or HG T1 disease. As there are significant implications in terms of CSS and OS with HG T1 vs. LG Ta, appropriate classification is key. Centralization of care may help with this.

Limitations / Discussion Points:
1. NCDB and SEER may overlap in terms of patients included. 

Presented by: Brian Jordan, Northwestern University, Chicago, Illinois
Co-Authors: Adam Weiner, Joshua Meeks, Chicago, IL

Written by: Thenappan Chandrasekar, MD, Clinical Fellow, University of Toronto, @tchandra_uromd at the 2018 AUA Annual Meeting - May 18 - 21, 2018 – San Francisco, CA USA