Urothelial carcinoma is a malignancy that originates in the genitourinary tract. It is a heterogeneous disease that can present at different stages, and the treatment options vary in efficacy. Advances in immunotherapy stimulated adoption in urothelial carcinoma, and published trials have shown promising results when compared to conventional therapies. However, oncologic drugs are historically costly, and immunotherapy is no exception. A cost-effectiveness analysis is a standardized method of weighing the clinical benefits of an intervention against the financial burden to obtain a composite proposed value. Multiple investigators have assessed immunotherapy in urothelial carcinoma, but no consensus has been reached.
1. Introduction
Urothelial cancer (UC) is a commonly diagnosed malignancy, with an estimated 550,000 new cases diagnosed worldwide in 2018
[1]. UC presents with a wide spectrum of disease stages and presentations, ranging from small, non-invasive, low-grade papillary tumors to diffuse metastatic disease. Treatment options vary depending on the grade and extent of tumor invasion. Certain treatments can be effective for oncologic control; however, there is an unmet need for treatments that provide a durable cure. For example, non-muscle invasive (NMI) carcinoma in situ (CIS) of the bladder harbors a 66% risk of progression to muscle-invasive (MI) disease, which intravesical therapy only partially mitigates to 10–20%
[2]. Similarly, most patients on first-line treatment for metastatic UC will ultimately progress. National committee guidelines still resort to recommending clinical trial enrollment for this population as even with modern therapies, the median survival remains less than 2 years
[3][4].
Immune checkpoint inhibitors (ICIs) emerged from the recognition that tumor cell evasion of the host response can occur by mimicking inhibitory signaling of healthy cells
[5]. The three most evaluated checkpoint targets with FDA approval are programmed cell death protein-1 (PD-1), programmed death ligand-1 (PD-L1), and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4). The first approved ICI for the treatment of UC was atezolizumab, which received accelerated approval in May 2016 (ultimately rescinded in 2021). Thereafter, the PD-L1 inhibitor avelumab followed by PD-1 inhibitors nivolumab and pembrolizumab were introduced. The CTLA-4 inhibitor ipilimumab has been evaluated in UC alone and in combination with nivolumab but has not yet received regulatory approval.
Multiple clinical trials have evaluated the use of ICIs for the treatment of various UC disease states, including the following notable examples. KEYNOTE-057 evaluated pembrolizumab for NMI Bacillus Calmette-Guerrin (BCG)-unresponsive CIS in a single-arm phase two study
[6]. Forty-one percent had complete response (CR) at three months with a median duration of CR of 16 months. For patients with muscle-invasive disease, the PURE-01 trial evaluated the efficacy of neoadjuvant pembrolizumab
[7]. Pathologic CR at the time of radical cystectomy (RC) was 42%, which improved when stratified by a combined positive score (CPS) ≥10% (a ratio of tumor and immune cells expressing PD-L1 to all tumor cells). Atezolizumab and nivolumab have also been explored in the neoadjuvant setting. Several ongoing adjuvant trials with anti-PD-1 therapies are underway, with at least one reporting a recurrence-free survival benefit
[8]. KEYNOTE-052 explored pembrolizumab as a first-line treatment for locally advanced or metastatic disease
[9]. A 28% response rate was noted with a median treatment duration of 30 months. Overall survival (OS) increased from 11 to 18 months when stratified by PD-L1 expression. KEYNOTE-045 investigated pembrolizumab as a second-line treatment for metastatic disease and recently updated results reaffirmed an improvement in the response rate and OS compared to chemotherapy
[10]. The above examples highlight the dynamic role of ICIs in UC as they continue to be investigated.
As with any new intervention, the improvement in outcomes must be balanced with the added cost. Cost-effectiveness analysis (CEA) is a standardized method of reporting and comparing interventions to better inform governing bodies on their added value for appropriate resource allocation. A new therapy with a large clinical benefit and minimal cost is more likely to be adopted than an expensive one with marginal benefit. CEAs are applied throughout healthcare but remain particularly relevant in oncology, where innovation constantly challenges standard therapy and costs are high compared to benign disease. This is supported by increased incorporation into disease management algorithms as demonstrated by the American Society of Clinical Oncology (ASCO) Value Network initiative. Public payors, such as the UK National Health Service, already have an established history of utilizing CEAs to determine medical coverage. The growing constraints on healthcare resources in the United States and abroad will only further the adoption of CEAs into health policy to maximize outcomes and minimize costs.
2. Cost-Effectiveness of Immune Checkpoint Inhibitors in Urothelial Carcinoma
The relatively recent application of ICIs to urothelial malignancies has continued to show promise as an alternative therapeutic pathway to current management. However, the potential improvement in clinical efficacy must be balanced with the additional cost. The growing importance of cost is well-illustrated by the National Institute for Health and Care Excellence’s (NICE) recent decision to reject coverage of pembrolizumab as it failed to meet their extended end-of-life threshold of £50,000/QALY in the UK
[11]. Here, we evaluated eight CEAs categorized by disease stage to highlight the potential for added value at each level.
On quality assessment, each study scored at least 18 of 20 possible points for items considered critical to include on economic evaluations. There is no specific cutoff for “good” or “poor” quality, but the authors of the CHEC list utilized here sought for investigators to attempt to meet each of the 20 unique points when executing cost-based studies
[12]. The only points missed on any study were having a relevant disclosure and/or not including a subjective ethics discussion. Disclosures are important as they represent potential conflicts of interest that can certainly introduce bias, but only one of four studies had a relevant disclosure that was aligned with the findings. While an ethics discussion is important, particularly in regard to value-based care, the absence does not adversely affect trial design or outcomes. Ultimately, each study included here met nearly all criteria considered critical to economic evaluation reporting.
The current best practice management of non-muscle invasive bladder cancer varies by risk category. The higher risk categories introduce a more complex clinical dilemma as higher risk of progression and ultimately metastasis must be weighed against the morbidity of cystectomy. KEYNOTE-057 attempted to complement intravesical therapy by evaluating pembrolizumab in BCG-unresponsive NMI UC. The data set the stage for the analysis by Wymer et al., which aptly included the appropriate disease states in their model and separated RC-eligible and ineligible patients into two independently analyzed cohorts. Pembrolizumab was, however, unable to meet the ICER threshold by a wide margin. It is difficult to envision pembrolizumab becoming cost-effective here in the near future as assumptions in the model already favored the ICI (i.e., no progression to metastasis in any patient) such that there would need to be a drastic improvement in outcomes data for a significant impact. Similarly, a > 90% price reduction was needed to meet the threshold. This is a larger discount than is typically seen even with conversion from patented to generic drug pricing. The main limitation of the study was the extrapolation of two-year outcomes to a five-year horizon, a common finding amongst the CEAs reviewed. Alternative approaches to reach cost-effectiveness may be investigating high-risk papillary disease instead of CIS as it may be more sensitive to treatment as well as trialing less expensive ICIs. Ultimately, radical cystectomy established itself as a higher value approach from a payor perspective and despite the morbidity will likely remain so until lower-cost alternatives are identified.
MI UC is encountered on presentation or after progression of NMI disease. The current recommendation is for RC, with consideration of trimodal therapy (TMT) for patients ineligible for surgery. The addition of neoadjuvant chemotherapy has repeatedly shown improvement in downstaging at RC as well as overall survival in this population
[13]. Grossman et al. demonstrated a five-year OS of 57% for a neoadjuvant MVAC cohort compared to 43% for RC only, with a higher pathologic CR at the time of RC of 38% vs. 15%
[13]. Dash et al. retrospectively compared GC to MVAC and found a similar pathologic CR rate of 26% and 28%, respectively
[14]. In both studies, lower pathologic stage at the time of cystectomy was associated with increased overall survival, suggesting similar efficacy of the two agents but with potential lower toxicity in the GC group. The pursuit of an effective neoadjuvant agent with limited secondary toxicity paved the way for ICIs.
In the single-arm phase II PURE-01 trial, patients with MIBC received three cycles of pembrolizumab followed by RC
[7]. Overall, 42% achieved pathologic CR at the time of RC with a 2% grade 3 or 4 AE rate. When stratified by tumor PD-L1 expression, 54% of patients with CPS ≥10% demonstrated CR vs. 13% in CPS <10%. The ABACUS trial was a similar phase II single-arm investigation into neoadjuvant atezolizumab
[15]. The pathologic CR rate at the time of RC was 31%, with 12% experiencing treatment-related grade 3 or 4 toxicity. Although there was a higher percentage of CR noted in the PD-L1-positive cohort (37%), this was not statistically significant. Van Dijk and colleagues evaluated a combination of nivolumab and ipilimumab in stage III UCC patients and demonstrated a 46% pathologic CR, with 55% of participants experiencing a grade 3 or 4 adverse event
[16]. The higher observed AE rate may have been partly due to the more advanced disease in this cohort.
Current accepted guidelines for advanced or metastatic UCC are first-line GC or MVAC, and carboplatin for cisplatin-ineligible patients. Second-line treatment is less concrete and varies from vinflunine, taxanes, to clinical trial enrollment. The lack of robust data in favor of a particular treatment has made it comparatively ripe with studies into the role of ICIs. In KEYNOTE-045, the pembrolizumab cohort had a statistically significant higher OS of 10.3 months from 7.4 months and the difference was maintained in tumors expressing PD-L1
[10]. CheckMate 032 studied nivolumab with and without ipilimumab in a similar population and demonstrated an objective response rate between 26% and 38% with combination therapy
[17]. Although IMVigor210 was encouraging for atezolizumab, the most recent update in IMVigor211 failed to meet the primary endpoint and FDA approval has since been rescinded
[18].
Second-line treatment was analyzed in four studies, which were split in their ability to meet WTP thresholds. Parmar et al. extrapolated from IMVigor211 and failed to find atezolizumab cost-effective. The lower rate of AEs and longer duration of response for atezolizumab were not enough to make up for the small difference in overall survival and large cost differential. This study was inherently limited as the trial it was based upon did not demonstrate an OS benefit. Slater et al. and Srivastava et al. deemed pembrolizumab cost-effective compared to conventional chemotherapy using KEYNOTE-045 results extrapolated to 20-year and 15-year horizons, respectively. The potentially higher durability of ICI response may indeed result in cost-savings, but without robust long-term survival data, the inferred outcomes can lead to overestimation of benefit. Parmar et al. demonstrate this effect as cost-effectiveness improved when increasing the 2-year trial follow-up to an extrapolated 10-year period. This is further supported by a retrospective review of various CEAs that demonstrated more favorable ICERs were calculated when longer time horizons were used (>5 years)
[19]. The pattern held true in the individual studies evaluating both short- and long-term horizons. One must be cautious accepting the results of extrapolated time horizons because a small survival benefit becomes artificially magnified without supporting real-world data. This is particularly relevant in the second-line treatment population where the life expectancy is significantly lower than other UC disease states.
Comparisons between the CEAs were limited by the heterogeneity in methodology. Studies included were preliminarily limited to those with ICERs to attempt standardization. Yet, variations persisted that made it difficult to perform head-to-head comparisons, such as the modeling technique, annual discount rate, and inclusion of additional costs, such as routine disease management or end-of-life care. Some studies extrapolated both treatment arms from the same trial, whereas others performed indirect treatment comparisons requiring an additional set of assumptions. The definition of the WTP threshold was also variable. This is a subjective cut-off that assigns a monetary value to a social good (i.e., $100,000 per QALY for pembrolizumab). In a review by Nanavaty et al., selection of a WTP value was influenced by various factors
[20]. The severity of disease mattered as diabetes treatments tended to have a more stringent limit then non-small cell lung cancer. Similarly, patients were afforded more leniency for end-of-life care. Certain countries held lower thresholds than others for the same intervention. Taken together, there is a subjective foundation to this mode of objective assessment. The subjectivity, while posing a limitation to quantitative analysis, appropriately reflects varying views on cost-effectiveness and value. These issues remain complex and may vary from one society to another, as clearly reflected by differences in health policy among countries.
3. Conclusions
UC is a heterogeneous disease with a need for more durable treatments. The introduction of ICIs into UC has encouraged a wave of promising outcomes-based studies, which need to be balanced with the higher associated costs. This has secondarily led to CEAs that attempt to gauge the value provided by these new immunotherapies. Though some analyses fell short of the set WTP threshold, the durable response of ICIs and favorable results in certain cohorts demonstrated a trend toward cost-effectiveness when longer time horizons and selective treatment algorithms were applied. There were considerable differences between studies, including the trial data used, the methodology employed, and the definition of value. This limited direct comparison among CEAs. The summarized data shows ICIs are not currently cost-effective in the treatment of UC, regardless of stage. However, improved patient selection, especially with the identification of a predictive biomarker, long-term outcome data, and increased competition leading to more favorable drug pricing, may improve value in the future.