Authors: Daniel A. Goldstein, Gary M. Ginsberg, Dorit Hofnung-Gabbay, Richard De Abreu Lourenco, Herbert H. Loong, Boon Cher Goh, Kelvin K. W. Chan, Massimo Di Maio, Francesco Perrone, Peter S. Hall, Alona Zer, Eli Rosenbaum
Categories: Original Investigation, Research, Online Only, Health Policy
Source: JAMA Network Open
Is extended-interval fixed dosing of pembrolizumab associated with increased health care costs?
This economic evaluation including simulated patients found that using the US Food and Drug Administration–labeled pembrolizumab dose of 400 mg every 6 weeks instead of 200 mg every 3 weeks may result in an estimated 8% increase in health care costs in the US.
These findings suggest that when considering different dosing options, clinicians and policy makers should consider the potential impact on costs.
As of January 2021, pembrolizumab has received 39 US Food and Drug Administration (FDA) approvals for the treatment of cancer. Initially approved with weight-based dosing, there was a subsequent movement toward using fixed doses for all patients, first with 200 mg every 3 weeks, and more recently 400 mg every 6 weeks.^1^ It has previously been argued that the move to fixed dosing from weight-based dosing has maintained efficacy but increased costs.^2^ While lower doses of immune checkpoint inhibitors have been strongly recommended in other publications,^3,4,5^ we recognize that the FDA-approved dose levels and schedules remain the standard of care in many health care settings.
As the dose intensity and price of pembrolizumab have essentially remained constant, it has been suggested that the cost of 400 mg every 6 weeks is equivalent to 200 mg every 3 weeks. We hypothesized that 400 mg every 6 weeks in fact represents a cost increase for health care payers. In clinical practice, when a patient’s disease progresses, the treatment is usually stopped irrespective of when the last treatment was received. However, if a patient is treated with larger, but less frequent doses, some of the drug may remain in the bloodstream, which would not otherwise have been infused if they had been receiving more frequent treatment at lower doses. The frequency of imaging studies to determine progression status will affect the costs associated with the duration of treatment. It is crucial to emphasize that a large body of literature exists to highlight that there are no expected differences in efficacy or safety of these 2 dosing strategies.^6^
We hypothesized that 400 mg every 6 weeks would be costlier for health care payers than 200 mg every 3 weeks for patients receiving pembrolizumab as second-line treatment for metastatic urothelial cancer. The primary aim of this study was to estimate whether extended interval dosing was more costly and, if so, to quantify this cost increase. The secondary aim was to assess the influence on the results of different drug and health care costs around the world. To examine this hypothesis, we used a cost minimization analysis to simulate the number of infusions, imaging studies, and drug consumption for various imaging frequency protocols for patients with metastatic urothelial cancer. We chose urothelial cancer because at the time of model development, pembrolizumab was a commonly accepted standard of care, and full survival data have been published.
This economic evaluation used a pharmacoeconomic model without the use of individual patient data; therefore, the study was exempt from the need for ethical approval or informed consent. This study follows the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) reporting guideline.
We used an Excel-based mathematical model (Microsoft) to simulate a population of patients receiving pembrolizumab as second-line treatment of metastatic urothelial cancer. The model ran from 0 to 104 weeks. Treatment was stopped at 2 years (104 weeks), in accordance with the registration trial protocol design, and common reimbursement restrictions. We estimated the duration of treatment until the medical decision to stop treatment. We incorporated data on drug and infusion costs. Our focus was to estimate the difference in treatment costs associated with the 2 dosing strategies for fixed dosing 200 mg every 3 weeks vs 400 mg every 6 weeks.
We simulated individual patients and extracted the progression-free survival (PFS) data of the pembrolizumab group from the registration clinical trial (KEYNOTE-045) to simulate the duration of treatment for each patient.^7^ For scenarios in which computed tomography (CT) imaging was performed in the same week as in the clinical trial, we used the trial’s PFS data. Since the largest decreases in the PFS data were only seen at the time of imaging, we interpolated these decreases by assuming that there was an underlying equal gradual PFS trend between each imaging event. These underlying PFS trend data were used for scenarios when CT imaging was performed in a week that had not been performed in the clinical trial. Further details of the modeling methods are included in the eMethods in Supplement 1.
The median duration of treatment in the KEYNOTE-045 trial was 3.5 months, and the median duration of treatment in our model was 3.3 months, closely replicating the median PFS in KEYNOTE-045. As in KEYNOTE-045, our model included a group of patients with durable responses to pembrolizumab. In KEYNOTE-045, the 12-month PFS was 17%, and this was exactly replicated in the model.
To calculate treatment costs incurred in the US, we used a commonly used method described by Tumeh et al.^8^ This method incorporates all direct costs from the hospital to the Centers for Medicare & Medicaid Services (CMS). Although there may be some variation in fees to private insurers, given that treatment for most patients with cancer is funded by CMS, we believe that this provides a close approximation. We used the June 2020 average sales price of 50.02 per mg of pembrolizumab.^9^ Treatment costs of 142.55 were based on a 1-hour infusion (Common Procedural Terminology [CPT] code 96413) according to the 2020 Medicare physician fee schedule.^10^ CT scan costs of $492.99 were based on the sum of an abdomen and pelvic scan with contrast (CPT code 74177) and a chest CT scan without contrast (CPT code 71250). As there was not expected to be any difference in adverse events rates (AEs) between the 2 groups of the model, this model did not consider the cost of AEs. Costs and consequences were evaluated over the full 104-week time horizon of the study to capture all the effects of differential dosing strategies, which ceased per the trial protocol after 104 weeks.
We performed cross-country comparisons using country-specific input values for drug and infusion costs. Recognizing that drug efficacy was expected to be constant across countries, we assumed that the duration of treatment would also be constant across countries. Country-specific drug and infusion costs were based on publicly listed prices, which did not incorporate any additional confidential negotiated discounts between the manufacturers and health care systems. Local currencies were converted to US dollars using the exchange rates as of mid-2020.
There are different clinical approaches to the frequency of required imaging, which may impact the time point of clinical decision-making and thus the duration of therapy. While some clinicians may perform imaging at different intervals, such as every 9 or 12 weeks, there may be other significant variations in imaging frequency in the clinical setting. This may be partially driven by payer approval policies or imaging capacity constraints. We therefore performed separate simulations to assess the potential impact of 3 different imaging strategies on overall costs in the model.
In the base case imaging scenario, we assumed that imaging started at week 9 and was repeated every 9 weeks thereafter, up to and including week 99 (ie, imaging at weeks 9, 18, 27, and so on). In the second imaging scenario, we assumed that imaging started at week 12 and was repeated every 12 weeks thereafter, up to and including week 96 (ie, at weeks 12, 24, 36, and so on). In the third imaging scenario, we assumed that imaging started at week 9 and was repeated every 6 weeks for the first year and subsequently every 12 weeks, up to and including week 99 (ie, at weeks 9, 15, 21, 27, 33, 39, 45, 51, 63, 75, 87, and 99).
Based on informal surveys of medical oncologists in different health care settings around the world, we believe that imaging scenario 1 is a common option in clinical practice; however, we recognize that scenario 2 may also be common. Scenario 3 was included because this was the imaging protocol used in the KEYNOTE-045 trial.^7^ Performing imaging every 12 weeks in scenario 2 will always be in sync with 6-week treatments, thus providing a low estimate of drug waste. By contrast, in scenario 3, imaging will always be out of sync with 6-week treatments, thus providing a high estimate of waste. In scenario 1, imaging will alternate between being in sync and out of sync with 6-week treatments, thus providing an estimate between the estimates of scenarios 2 and 3. In essence, when imaging and infusions are in sync, there should be no waste, and when they are completely out of sync, it enables us to calculate the maximum possible waste.
We hypothesized that dosing every 3 weeks would be cheaper in terms of drug costs alone than dosing every 6 weeks. However, recognizing that infusion costs will be higher for the more frequent treatment strategy, we suspected that there may be a transition point for patients with durable responses at which the less frequent dosing becomes less costly due to lower administration costs. We sought to identify whether and when this transition point occurs. Therefore, assuming imaging scenario 1, further sensitivity analyses were performed in which infusions were initially given every 3 weeks but transitioned to every 6 weeks at week 30, 42, 54, or 84.
Some health care systems around the world use a weight-based dosing approach, using 2 mg/kg every 3 weeks or 4 mg/kg every 6 weeks.^11^ Given that the mean weight of a patient with cancer is 75 kg, we performed a sensitivity analysis using mean doses of 150 mg every 3 weeks compared with 300 mg every 6 weeks.
While we recognize that for the health care payer, the differences in infusion costs are relevant, they are not relevant for the manufacturer. We therefore assessed the overall costs and revenue from both of these different perspectives.
Based on first quarter reports in 2020, global pembrolizumab sales in 2020 were estimated to be approximately 7.3 billion) were in the US.^13^ According to the manufacturer, urothelial cancer accounted for 5% of all sales.^14^ Therefore, 2020 expenditure on pembrolizumab in the US is estimated to have been approximately $367 million. We sought to estimate the overall budget impact of the 3- and 6-week dosing strategies in the US under different imaging scenarios, modified by the application of an annual discount rate of 3%^15^ applied to all costs incurred during the second year of the evaluation. Data were analyzed from 2020 to 2022.
In the base case imaging scenario 1, we estimate that dosing every 6 weeks instead of every 3 weeks would result in an 8.9% increase in pembrolizumab costs for the health care payer (Table 1, Figure 1). This strategy would result in an annual increase in manufacturers revenues in the US of approximately 28 million (Figure 1). Figure 2 illustrates mixed dosing strategies. If the transition to the 6-week dosing schedule occurred at week 30, total cost increases would be reduced to 1.5%, and transition at week 42 would result in an increase of 1.8% (eTable 2 in Supplement 1). No additional significant cost differences would occur if the transition occurred at 54 or 84 weeks.
Figure 1. Estimated Extra Annual Costs of 6-Week Dosing Schedule in the US From Manufacturers and Payers’ Perspectives
Figure 2. Estimated Costs of Mixed Dosing Strategy Using Base Case Imaging Scenario 1Week of transition is when patients are transitioned from a 3-week to 6-week dosing schedule.
In imaging scenario 2, we estimate that dosing every 6 weeks instead of every 3 weeks would result in a 3.3% increase in pembrolizumab costs for the health care payer (Table 1, Figure 1). This strategy would result in an annual increase in manufacturer revenues in the US of approximately 7.5 million (Figure 1).
In imaging scenario 3, we estimate that dosing every 6 weeks instead of every 3 weeks would result in a 9.4% increase in pembrolizumab costs for the health care payer (Table 1, Figure 1). This strategy would result in an annual increase in manufacturers revenues in the US of approximately 29 million (Figure 1).
Table 2 and Figure 3 show unit drug and infusion costs for selected countries. As a result of moving from a 3-week to 6-week dosing protocol, additional drug costs per patient would range from 8072 in the US. After considering the cost savings resulting from fewer infusions, the net cost increase for health care payers range from 7483 in the US (Table 2). In an additional sensitivity analysis on the base case using mean weight-based dosing, we estimate that 300 mg every 6 weeks instead of 150 mg every 3 weeks would result in an 8.5% net cost increase, or $31 million, from the US payers’ perspective.
Figure 3. Estimated Mean Additional Costs Using 6-Week Dosing Schedule Instead of 3-Week Dosing Schedule, Per Patient and by Country Costs were modeled with baseline imaging scenario 1.
This economic evaluation estimates the potential financial impact of extended-interval dosing of cancer drugs. We used the example of pembrolizumab in metastatic urothelial cancer as an example to demonstrate this issue; however, this concept is certain to be applicable to other cancers using the same drug as well as other drugs. To our knowledge, this is the first pharmacoeconomic study to demonstrate this issue and to provide the modeling method. Furthermore, we demonstrated how the model could be used and that it was robust when applied to health care systems in different countries.
It is important to highlight that, in our model's estimates, increased pharmaceutical costs (as a result of moving to a 6-week dosing protocol) to health care payers would be only partially ameliorated by resultant savings in infusion costs. From the manufacturer’s perspective, replacing 3-week dosing with 6-week dosing led to an increase in sales revenue in all possible scenarios. While our model demonstrated potential financial impact to manufacturers and health care payers, it is fair to assume that there may be some impact on individual patients, depending on which country they live in and how their health care is financed. Clearly, there is also a benefit to 6-week dosing, namely patient convenience due to fewer visits to the hospital and potentially also less exposure to infections from hospital visits. The question for policy makers is whether the increased cost is justified by the increased convenience. Our study provides financial data, thus helping policy makers to more fully consider this trade-off.
This study has some limitations. As with all models, there are limitations based on estimations used in the model inputs. The precise treatment duration was estimated using PFS as a surrogate. While we believe this to be a good surrogate, it may not be perfectly accurate. Furthermore, in the KEYNOTE-045 trial, there were some patients who continued therapy beyond progression for unclear reasons. We did not incorporate this phenomenon into the model, as we suspect that this would not be common practice in the practice setting. Using the clinical trial as the basis for the model may also be problematic in some respects, as clinical trials are often not fully representative of the practice setting. There may be intracountry and intercountry differences in patterns of care due to socioeconomic diversity, which may lead to variations in treatment durations. We used cost inputs that were publicly available; however, there may be subsequent confidential discounts provided to health care payers. There may also be intracountry price differences; however, we tried to use price estimates that were most representative of the overall population in each country. In estimating drug and infusion costs, there is some variation among countries. While some variation was expected, the precise reasons for variation and its magnitude are unclear. We used second-line pembrolizumab as the example in our model; however, we recognize that the use of this regimen in this setting may decrease with time due to the recent publication of first-line switch maintenance with avelumab.^16^ While not all patients will receive maintenance avelumab, as the uptake increases, the use of second-line pembrolizumab will inevitably decrease. This will not affect the percentage increased cost of using extended interval dosing, but it will reduce the total increase in expenditure. While this model design can be replicated for other malignant neoplasms, differing drug efficacy, patterns of ongoing monitoring, and treatment durations may lead to different results. We recognize that due to the unfavorable prognosis in the setting of second-line urothelial cancer, treatment interruptions may be more common than in the setting of other diseases. A limitation of this modeling study is that it uses set imaging frequency protocols, and does not account for some clinical situations of off-schedule imaging due to onset of new symptoms. Our clinical impression is that these situations would be uncommon, would be balanced in both groups of the model, and would not have a major impact on the model results.
There are of course many other issues that affect costs when using immunotherapy in cancer care, and in this economic evaluation, we are highlighting only one of these issues. Costs could be reduced by reducing while the current fixed dosing strategies appear on the FDA label, it is well established that weight-based dosing can provide the same efficacy while reducing costs.^3^ Such an approach has been incorporated by national health care agencies, such as in Canada.^11^ Furthermore, the need for prolonged therapy in responding patients is an open question. While most of the clinical trial protocols used therapy up to a maximum of 2 years, this may not be necessary.
In this economic evaluation model comparing different fixed dosing options of pembrolizumab, we found an approximately 8% estimated increase in health care costs when using the FDA-labeled dose of 400 mg every 6 weeks instead of 200 mg every 3 weeks. As new treatments and technologies are developed, health care payers need to continue to be mindful of cost implications prior to implementation. This study provides a new concept and analytic model for consideration.