Introduction:

Intravesical induction BCG once weekly for six weeks is the mainstay of initial treatment for intermediate or high risk non-muscle invasive bladder cancer (NMIBC). However, the value of ongoing maintenance BCG for one to three years is unclear. In a meta-analysis of 20 randomized clinical trials examining maintenance BCG, Sylvester et al found that maintenance BCG was associated with reduced bladder cancer recurrence and progression to MIBC. A subsequent patient level meta-analysis of BCG maintenance trials demonstrated reduced recurrence but no difference in progression with maintenance BCG. Given the unclear effect on progression, limited supply, and increased cost and toxicity of maintenance BCG, 50% of urologists do not use maintenance BCG for intermediate and high risk NMIBC patients.    

In a resource constrained health care infrastructure plagued by BCG shortages, it is critical to understand the value of maintenance BCG. Herein, we conduct a comprehensive cost-effectiveness analysis on maintenance BCG using Markov modeling. Such a model could provide valuable insight on balancing the costs and benefits of maintenance BCG. Furthermore, it could identify thresholds for reduction in both recurrence and progression at which maintenance BCG would be considered cost-effective.   

Methods:

A Markov model was constructed to compare the cost-effectiveness of maintenance BCG relative to no-maintenance BCG after successful completion of induction BCG for intermediate and high risk NMIBC. A simplified model schematic is shown in Figure 1. Analysis was from a US payer perspective using Medicare costs, and utility values were obtained from the literature. The following maintenance BCG schedule was followed: once weekly instillations for 3 weeks at 3, 6, 12, 18, 24, 30, and 36 months after induction BCG completion. Intermediate risk patients were scheduled to receive 1 year of maintenance BCG and high risk patients 3 years. Surveillance consisted of guideline based cystoscopy intervals and annual CT urograms in both the maintenance and no-maintenance BCG arms of the model. Maintenance BCG side effect and drop-out rates were taken from large randomized trials. 5-year recurrence, progression to MIBC, metastasis, and mortality rates were extracted from meta-analyses and randomized trials. In the base case, maintenance BCG was associated with 20% improved 5-year recurrence free survival but not progression free survival. We then identified a threshold of reduction in recurrence alone below which maintenance BCG would be cost-effective. Next we identified a threshold of reduction in progression below which maintenance BCG would be cost effective. Other univariable sensitivity analyses were performed for all relevant variables in our model. Multivariable sensitivity analyses were performed using 10,000 Monte-Carlo microsimulations at a willingness-to-pay threshold of $100,000 per quality adjusted life year (QALY).

Results:

At 5 years after randomization, mean costs and QALYs per patient were $10,069 and 2.73 QALYs for maintenance BCG and $8,901 and 2.73 QALYs for no-maintenance BCG, respectively. In the base model, assuming a 20% 5-year recurrence free survival benefit for maintenance BCG and no progression free survival benefit, the excess costs of maintenance BCG and the associated lower utility from BCG related side effects balanced out the costs and lower utility associated from increased recurrence in the no-maintenance BCG arm. On sensitivity analysis, maintenance BCG became cost-effective using a willingness-to-pay threshold of $100,000/QALY if its absolute 5 year recurrence rate was 24% lower than no-maintenance BCG (assuming no difference in MIBC progression) or if the 5 year progression rate to MIBC was 2.4% lower for maintenance BCG vs no-maintenance BCG. Figure 2 displays a Tornado diagram to demonstrate the impact of varying individual model variables on the net-monetary benefit of maintenance BCG, in which variables that significantly altered maintenance BCG cost-effectiveness are marked with a vertical black bar. In addition to recurrence and progression, maintenance BCG became cost-effectiveness if its 5-year overall mortality rate was 0.5% lower than no-maintenance BCG. On multivariable sensitivity analysis using 10,000 Monte-Carlo microsimulations of the base case (20% difference in 5 year recurrence and no difference in progression or mortality), maintenance BCG was cost-effective in 22% of microsimulations at a willingness-to-pay threshold of $100,000/QALY. 

Conclusion:

Maintenance BCG has a low chance of being cost-effective for intermediate/high risk NMIBC at traditional willingness-to-pay thresholds assuming a 20% reduction in 5-year recurrence and no difference in progression to MIBC relative to no-maintenance BCG. Maintenance BCG became cost-effective if it reduced 5-year recurrence rates by 24% or more, and thus efforts to sub-stratify the impact of maintenance BCG on recurrence for intermediate and high risk NMIBC are warranted to improve patient selection for maintenance BCG.    

Maintenance BCG also became cost-effective if it was associated with a 2.4% 5-year progression-free survival benefit. A traditional meta-analysis found a statistically significant 4% 5-year progression benefit (driven largely by an old trial in which repeat transurethral resection for high risk NMIBC was not employed) but an individual patient-data meta-analysis found no significant difference in progression. Given this ambiguity, our model supports further high quality studies that clarify the impact of maintenance BCG on progression to MIBC for intermediate and high risk NMIBC.    

In the present milieu of chronic BCG shortages, our study supports the American Urologic Association’s current position of prioritizing BCG for induction courses over maintenance regimens. 

Funding: N/A