As the therapeutic options for patients with nonmuscle invasive bladder cancer (NMIBC) expand,¹ choosing the right treatment at the right time for the right patient becomes increasingly challenging. Providers must consider NMIBC risk stratification, the reported efficacy of potential treatments, side-effect profiles, costs, and logistical concerns such as the burden of scheduling, as well as patient goals. Frequent bacillus Calmette-Guérin (BCG) supply shortages and the lack of clinical trials directly comparing novel therapies further complicate decision making.

Cost-effectiveness analysis (CEA) represents a relatively underutilized yet powerful technique that has the potential to improve care for NMIBC patients. While commonly used to comment on the pricing of interventions, cost-effectiveness analyses can more importantly integrate the spectrum of clinically relevant information to inform guideline recommendations and treatment decisions for NMIBC.

Cost-effectiveness analyses–or, more appropriately, cost-utility analyses–assimilate information on the quality and quantity of life as well as cost for each treatment option. The Appendix outlines basic definitions of CEA terminology.² A common CEA format utilizes Markov models with relevant “health states” downstream of each treatment option. The figure is a schematic of sample NMIBC health states such as recurrence, progression, and toxicity. Each health state is ascribed unique costs and a utility value, which reflects the quality of life for the given state. The probability of transitioning between health states for each time cycle in the Markov Model is derived from oncologic and toxicity data from the literature. The utility values for all health states downstream of a treatment are then summed in a probability-weighted manner for every time cycle to calculate the quality-adjusted life-years (QALYs) for the given treatment. A similar summation is performed for costs. The “expected” QALYs and costs for treatments can then be compared using the Incremental Cost-Effectiveness Ratio (ICER), which is the difference in cost divided by the difference in QALYs between 2 treatments. Treatments with an ICER of less than $100,000 per additional QALY (the conventional willingness to pay threshold) are considered cost-effective.

Figure. Hierarchy illustrating several of common health states typically experienced by patients with NMIBC. Note: this representation is not intended to be a Markov Diagram demonstrating transitions possible from one health state to another.

Although this approach may seem abstract, the resulting models are particularly useful in NMIBC decision making. For example, given frequent BCG shortages, and the resulting emphasis on conservation and appropriate allocation of supply,³ we conducted a CEA evaluating maintenance BCG for intermediate and high risk NMIBC.⁴ We found that maintenance BCG only became cost-effective if it reduced 5-year progression by at least 2.1%. Given the modest effect size of maintenance BCG on progression, we reasoned that maintenance BCG is likely only cost-effective for subsets of high risk NMIBC with a significant risk of 5-year progression and can likely be foregone for intermediate risk NMIBC, especially during times of BCG shortage. In fact, further probabilistic sensitivity analysis demonstrated that maintenance BCG was only cost-effective in ∼17% of intermediate and high risk NMIBC simulations. In this way, the CEA not only identified which patients would benefit most from maintenance BCG treatments, but also aligned with American Urological Association policy statements on the matter.⁵

Moreover, determining treatment for patients with BCG-unresponsive NMIBC represents another significant opportunity for CEA utilization. Here, the decision making remains challenging as the U.S. Food and Drug Administration has permitted single-arm trials to be used for agent approval,⁶ causing the number of agents being tested in this disease space to increase rapidly. CEAs thereby offer the opportunity to contextualize the results of the existing and emerging single-arm trials to standards of care, such as radical cystectomy or salvage intravesical chemotherapy. As a proof of concept, we recently evaluated the cost-effectiveness of pembrolizumab relative to radical cystectomy and intravesical gemcitabine-docetaxel for patients with BCG-unresponsive carcinoma in situ.⁷ We found that pembrolizumab was not cost-effective relative to radical cystectomy (ICER $1,403,008) or intravesical gemcitabine-docetaxel (ICER $2,011,923), regardless of cystectomy eligibility status. Its marginal benefit (0.10 QALYs) over radical cystectomy was not justified by its high cost and high risk of adverse events. In fact, a greater than 90% price reduction would be required to render pembrolizumab cost-effective for this indication. However, we did find that intravesical gemcitabine-docetaxel, based on retrospective multi-institutional data,⁸ was cost-effective in about 46% of simulations vs radical cystectomy and warrants further study as a treatment option. Thus, by quantifying the balance of side-effects, oncologic benefit, and cost relative to existing standards of care, CEAs can help interpret single-arm trial data for BCG-unresponsive NMIBC and guide future treatment selection.

Other potential uses for CEAs in BCG-unresponsive NMIBC include testing the relative importance of various oncologic endpoints over complete response thresholds, which may improve trial design.⁶ In addition, CEAs may be incorporated into shared decision making tools (akin to WiserCare© in prostate cancer) that weigh treatment risks and benefits according to an individual patient’s preferences to help patients decide between treatment options. CEAs may also provide rough estimates of expected costs for patients, thereby assisting in patient counseling. Considering that NMIBC is one of the most expensive cancers on a per-patient level, CEAs can as well inform personalized surveillance schedules for NMIBC patients.

Importantly, CEAs are limited by the quality of data used for the modeling, including the oncologic and cost data sources. A particularly critical limitation is the utility value data, which can vary widely depending on the method used to derive the utility values⁹ and thus alter expected QALY calculations.¹⁰ Nevertheless, this limitation may be addressed by encouraging the collection of utility value assessments along with patient-reported outcome questionnaires in clinical trials. In addition, sensitivity analyses varying utility values over broad ranges can be conducted to determine how susceptible the model is to small changes in utility values (see table).

In summary, cost-effectiveness analyses offer the opportunity to quantify the costs, toxicity and oncologic benefit of NMIBC treatments, facilitating both an understanding of treatment tradeoffs and the comparison of novel agents to existing standards of care. Sensitivity analyses may further identify subsets of patients for whom a given treatment may be preferred, thereby enabling more personalized treatment recommendations. In this manner, CEA represents a tool to assist clinicians as well as guideline panels on interpreting data and individualizing options for NMIBC patients.