Conclusions

The pharmacoeconomic results are informed by results from the GARNET study, a single-arm, phase I trial that aimed to assess the efficacy and safety of dostarlimab among patients with deficient mismatch repair (dMMR) or microsatellite instability-high (MSI-H) recurrent or advanced endometrial cancer (EC). According to the CADTH clinical review, the lack of a comparator group precludes the ability to assess the relative therapeutic benefit or safety of dostarlimab to relevant comparators (i.e., active treatment, standard of care). According to the CADTH clinical review, it is not clear from the GARNET study whether dostarlimab may improve progression-free survival (PFS). There is indirect evidence to suggest that dostarlimab may improve PFS, compared with the standard of care (chemotherapeutic regimens); however, no strong conclusions can be drawn because of many limitations in the indirect comparisons. The relative effects of dostarlimab on patient-relevant outcomes, such as overall survival (OS) and quality of life, are uncertain.

CADTH was not able to conduct a CADTH base-case analysis because of foundational limitations with the sponsor’s model and submitted clinical evidence. Notably, the comparator population was not matched by dMMR or MSI-H status and, thus, the cost-effectiveness of dostarlimab compared with second-line treatments in the indicated population is unknown. The sponsor’s choice of a partitioned survival model (PSM) may overestimate incremental quality-adjusted life-years (QALYs), and the long-term extrapolations of OS and PFS are highly uncertain. CADTH conducted an exploratory analysis, the results of which suggest that the model is highly sensitive to several model assumptions, including long-term survival extrapolation, stopping rules, and treatment-effectiveness waning after 24 months.

The sponsor submitted a model comparing the cost-effectiveness of dostarlimab with second-line treatments (pegylated liposomal doxorubicin [PLD]; doxorubicin; paclitaxel; current mix of treatments [CMT]; carboplatin plus PLD; carboplatin plus paclitaxel; and carboplatin). According to the comparative evidence in the submitted model, carboplatin plus paclitaxel was the treatment with the smallest amount of uncertainty related to OS and PFS. Based on the sponsor’s submitted base case, the probability that dostarlimab is cost-effective was 0% at a willingness-to-pay threshold of $50,000 per QALY. The sponsor’s base-case results indicate that 89% of the incremental benefit of dostarlimab was obtained beyond the single-arm trial period, when cost-effectiveness results vary widely, depending on the chosen extrapolation function. Based on the CADTH exploratory analysis, a price reduction of 83% would be needed for dostarlimab to be considered cost-effective at a willingness-to-pay threshold of $50,000 per QALY; however, this estimate is subject to limitations in the sponsor’s submission, most crucially the inappropriateness of the estimated efficacy of comparator treatments in the dMMR or MSI-H population. An additional price reduction may be warranted.

Treatment of dMMR or MSI-H EC with dostarlimab increases costs, compared with other currently available treatments. CADTH could not estimate the amount of health improvement that may result from treatment with dostarlimab because of highly uncertain clinical efficacy data and a lack of information about the efficacy of comparator treatments. Because of methodological limitations identified in the model and clinical evidence, the cost-effectiveness of dostarlimab is unknown.

Economic Evaluation

Summary of Sponsor’s Economic Evaluation Results

All analyses were run probabilistically (5,000 iterations for the base-case and scenario analyses). The deterministic and probabilistic results were similar. The probabilistic findings are presented below.

Base-Case Results

The sponsor’s base-case results were calculated as pairwise comparisons and as sequential analyses. In the sponsor’s sequential analyses, the CMT comparator was excluded. The sponsor submitted pairwise comparisons between dostarlimab and each individual comparator treatment. CADTH has presented the pairwise comparison for dostarlimab versus carboplatin plus paclitaxel, as it is the treatment with the highest number of observed events (i.e., least uncertainty) and is the treatment option that is next on the cost-effectiveness frontier (i.e., smallest number of incremental QALYs). All submitted pairwise comparisons are presented in Appendix 3.

In the sponsor’s base-case results of the pairwise comparisons between dostarlimab and individual comparators, incremental costs associated with dostarlimab ranged from $804,503 (compared with carboplatin plus paclitaxel) to $829,598 (compared with doxorubicin). Incremental QALYs gained ranged from 4.79 for dostarlimab versus carboplatin to 5.93 for dostarlimab versus doxorubicin or PLD. The incremental cost-effectiveness ratio (ICER) for dostarlimab ranged from $138,486 per QALY gained versus PLD to $171,989 per QALY gained versus carboplatin. Compared with CMT, incremental QALYs was 5.12 for dostarlimab and incremental costs were $816,233, with an ICER of $159,352 per QALY gained.

Results were driven by OS projections that predicted substantial differences in total LYs and increased drug-acquisition costs associated with dostarlimab. The sponsor’s submitted ICERs ranged from $138,000 to $172,000 (details in Appendix 3). Based on the sponsor’s base case, approximately 89% of the incremental benefit for dostarlimab was derived from the extrapolated period (beyond the single-arm trial period).

Table 3

Summary of the Sponsor’s Economic Evaluation Results (Dostarlimab Versus Carboplatin + Paclitaxel).

Sensitivity and Scenario Analysis Results

The sponsor conducted several scenario and sensitivity analyses, which included various discount rates (0% and 3%); the adoption of alternative parametric curves for the extrapolation of OS (exponential and lognormal) and PFS (1-knot spline and 2-knot spline) for dostarlimab; the assumption that the OS efficacy of dostarlimab after treatment waning is equal to that of CMT starting 5 years after treatment initiation out to 10 years; the application of alternative data sources for OS hazard ratios of comparator therapies (i.e., Flatiron Health database for CMT¹; the ZoptEC clinical trial for doxorubicin¹¹; the McMeekin et al. [2015]¹² study of paclitaxel and docetaxel; and the Mazgani et al. [2008]¹³ study of carboplatin plus paclitaxel); removal of the cap for the number of treatment cycles for chemotherapies; exclusion of age-related disutilities; and exclusion of IV drug administration costs.

Several scenarios resulted in notable increases in the ICER, including the adoption of alternative parametric curves for the extrapolation of OS and PFS. Notably, the sponsor’s model was highly sensitive to the use of the exponential parametric distributions for OS for dostarlimab (resulting in an ICER that ranged from $218,206 per QALY gained versus PLD to $450,209 per QALY gained versus carboplatin plus paclitaxel), and the use of the lognormal parametric distribution, which resulted in an ICER that ranged from $199,122 per QALY gained versus PLD to $279,935 per QALY gained versus carboplatin plus paclitaxel.

CADTH Appraisal of the Sponsor’s Economic Evaluation

CADTH identified several key limitations to the sponsor’s analysis that have notable implications for the economic analysis:

• The comparative efficacy and safety of dostarlimab to relevant comparators is highly uncertain: A key limitation of the clinical efficacy (i.e., OS and PFS) data informing dostarlimab in the economic model is that it is based on the GARNET study, a nonrandomized, open-label, single-arm study without a comparator group. The primary objective of phase I (nonrandomized) studies is to document safety outcomes rather than efficacy. In the GARNET study, there were a limited number of patients in the efficacy dataset, and the interpretation of OS with dostarlimab is limited, owing to immature OS data and the short duration of follow-up. Additionally, the results for HRQoL and symptom-severity exploratory outcomes remained inconclusive, due to a number of important limitations, as identified by the CADTH clinical review. In the absence of comparative clinical evidence from the GARNET study to inform the efficacy (i.e., OS and PFS) and safety of dostarlimab, compared with comparator treatments, in the submitted model, and considering the lack of a common comparator, the sponsor undertook 6 indirect treatment comparisons (ITCs) — 3 MAIC reports and 3 IPTW reports — to assess the efficacy of dostarlimab compared with individual monotherapy and combination chemotherapy regimens. CADTH’s review of the clinical evidence identified a number of methodological limitations to the sponsor’s MAICs and IPTW analyses, which add considerable uncertainty to the cost-effectiveness analysis. Clinical experts consulted by CADTH indicated that in the absence of robust comparative data on PFS and OS, no firm conclusions could be drawn about how dostarlimab compares with other relevant treatment options.
  • Given the absence of direct evidence and the many limitations of the sponsor’s multiple ITCs, the cost-effectiveness of dostarlimab, compared with single-agent and compared with combination chemotherapies, is highly uncertain. The comparative effectiveness of dostarlimab versus relevant comparators remains highly uncertain.
• The patient population in the comparator group was not matched for dMMR or MSI-H status and did not reflect the indicated population: The sponsor derived data on the comparator group from patient-level health data available from NCRAS, a synthetic real-world cohort of patients with advanced or recurrent EC in the UK. Although patients in cohort A1 in the GARNET trial included those with dMMR or MSI-H EC, per the indication, information on dMMR or MSI-H status was not available in the real-world cohort database. According to the clinical experts consulted by CADTH, dMMR or MSI-H status is a prognostic factor and strongly influences clinical management. This limitation in the sponsor’s submitted evidence means that the efficacy of dostarlimab, compared with all comparator treatments, is unknown for a dMMR and MSI-H population.
  • CADTH could not address this misalignment of the patient population through reanalysis. The cost-effectiveness of dostarlimab relative to all comparator treatments is unknown.
• Model structure suggests a post-progression survival benefit: Results from the sponsor’s model suggest that dostarlimab is associated with longer survival after disease progression. Specifically, the sponsor’s results suggest a post-progression survival benefit for patients receiving dostarlimab, relative to carboplatin plus paclitaxel (the next-nearest comparator), such that roughly 42% of the incremental survival (2.84 LYs) would occur after patients have experienced disease progression and have discontinued dostarlimab (Table 13). The CADTH clinical review team noted that there is no evidence of a clear mechanism by which dostarlimab would continue to provide clinical benefit for patients with progressive disease, given that the GARNET study was a single-arm study and, therefore, a treatment effect between groups could not be assessed. The sponsor’s use of a PSM introduces structural assumptions about the relationship between PFS and OS that likely do not accurately reflect causal relationships within the disease pathway. These assumptions may produce a post-relapse survival bias that favours dostarlimab. Because of the structural independence between OS and PFS end points assumed in a PSM, extrapolations for each end point may reflect within-trial trends in the rates of progression and death. Last, the clinical experts consulted by CADTH observed that the results for carboplatin monotherapy yielded higher survival rates in this patient population than carboplatin plus docetaxel. This finding was considered implausible. Accordingly, CADTH omitted this finding from its analysis and chose the combination therapy as the comparator closest on the frontier to dostarlimab. For this reason, CADTH presented pairwise comparisons between dostarlimab and carboplatin plus paclitaxel.
  • CADTH asked the sponsor to provide additional evidence to support the implied post-progression benefit (2.84 incremental LYs and 1.98 incremental QALYs, compared with carboplatin plus paclitaxel); however, the sponsor did not provide any clinical evidence from the GARNET study to substantiate this claim. Rather, the sponsor asserted that RWE data suggested a post-progression benefit from dostarlimab would be possible because the proportion of patients likely to receive platinum-based therapy in the third-line would be higher than the proportion likely to receive currently available treatment options in the second-line; because a higher proportion of patients receiving dostarlimab in the model could receive active therapy after discontinuation; and because OS and PFS extrapolations applied in the sponsor’s base case represented the mechanism of action of a post-progression benefit for dostarlimab over time. The CADTH clinical review of the sponsor’s submitted evidence did not find evidence to support this explanation. The clinical experts consulted by CADTH noted that the sponsor’s chosen comparators did not match expected clinical practice and, therefore, did not support this explanation of the post-progression survival difference observed in the sponsor’s model. CADTH was unable to determine the extent to which the implied post-progression benefit was due to the effect of treatment with dostarlimab, structural bias in the PSM, or limitations in the comparator efficacy evidence.
• Long-term extrapolations of the clinical efficacy (OS and PFS) of dostarlimab are likely overestimated: The sponsor fitted several parametric survival curves to extrapolate OS and PFS for patients who received dostarlimab over the lifetime time horizon (40 years), based on the observed period of the GARNET trial (median duration of follow-up = 16.3 months). According to the clinical experts consulted by CADTH, the estimates of OS and PFS beyond the observed period used in the GARNET study were unrealistically high and did not align with the anticipated prognosis for this patient population. The CADTH clinical review did not find evidence in the GARNET study to support the sponsor’s survival extrapolations.
  The sponsor also fitted parametric survival curves for OS, PFS, and TTD for each individual comparator based on the UK RWE cohort data. The clinical experts consulted by CADTH indicated that the extrapolated OS and PFS estimates were overestimated and did not align with the natural history of Canadian patients seen in clinical practice. Based on the second-line therapies commonly used in Canadian clinical practice, the experts indicated that approximately 60% to 70% of patients would likely remain alive at 6 months, 30% at 12 months, and 10% or so at 24 months, followed by a rapid decline in OS and PFS.
  • Incremental QALYs for all comparators are likely overestimated as a result of this limitation. CADTH explored alternative assumptions in exploratory analyses (refer to Appendix 4).
• The sponsor’s choice of comparators may not reflect current standard-of-care treatments offered in Canadian clinical practice: The sponsor excluded several chemotherapies (monotherapies and combined treatment regimens) that were identified to be relevant comparators for adults with recurrent or advanced dMMR or MSI-H EC who have progressed on or after prior treatment with a platinum-containing regimen (i.e., pembrolizumab, cisplatin, carboplatin plus docetaxel, carboplatin plus doxorubicin, or cisplatin plus doxorubicin), according to the clinical experts consulted by CADTH. Among the comparator treatments that were included, the clinical experts consulted by CADTH noted that several did not reflect Canadian clinical practice. Of note, the experts stated that hormonal therapy is rarely used in the indicated population and, importantly, there is no true standard of care for second-line therapies.
  Additionally, the frequency of use of certain treatments as part of the standard of care in the second-line setting for the comparator group did not reflect Canadian clinical practice, particularly the distribution of patients who received a platinum doublet as a second-line therapy, such as carboplatin plus paclitaxel. The clinical experts consulted by CADTH noted that patients with dMMR or MSI-H disease are typically not sensitive to platinum-containing therapies and are therefore not eligible for re-treatment with platinum-containing regimens after progression. Accordingly, the frequency of use in the sponsor’s submission was not reflective of clinical practice for the indicated population. The clinical experts consulted by CADTH suggested that paclitaxel or liposomal doxorubicin would be the most frequently used comparators.
  • CADTH explored alternative assumptions in an exploratory analysis (refer to Appendix 4).
• Patients who received dostarlimab were assumed to remain on treatment indefinitely over the lifetime time horizon, and the duration of treatment effect was assumed to last for a patient’s lifetime, without any treatment waning: In the sponsor’s base case, the sponsor did not apply a stopping rule or any treatment discontinuation criteria to patients who received dostarlimab. As such, patients were assumed to continue receiving dostarlimab indefinitely over the lifetime time horizon (40 years). Additionally, the sponsor assumed that the duration of treatment effect for dostarlimab would last for the patient’s lifetime (i.e., no treatment waning). Each of these assumptions were noted to be unlikely by the clinical experts consulted by CADTH. First, the clinical experts consulted by CADTH noted that 50% to 60% of patients were likely to discontinue treatment after 2 years, and the majority (approximately 90%) would likely be off treatment by 5 years. Second, CADTH’s clinical experts indicated that treatment-waning effects were likely to occur approximately 2 years after treatment initiation, based on their clinical experience with patients receiving immunotherapy in Canadian clinical practice.
  • CADTH explored alternative assumptions in an exploratory analysis (refer to Appendix 4).
• Health state utility values for the progressed disease health state lacked face validity: In the economic model, health state utility values were estimated with EQ-5D-5L data collected from patients in the GARNET study. Based on the sponsor’s methodology, utility values for patients in the progressed disease health state (0.74) were similar to those for patients in the progression-free health state (0.78). The clinical experts consulted by CADTH indicated that patient quality of life typically worsens with disease progression. As such, health state utility values for the progressed disease health state lack face validity and likely overestimate patients’ post-progression quality of life in favour of dostarlimab. The magnitude of bias in favour of dostarlimab remains unknown, resulting in additional uncertainty about the impact of health state utility values on the ICER.
  • CADTH explored alternative assumptions in an exploratory analysis (refer to Appendix 4).

Table 4

Key Assumptions of the Submitted Economic Evaluation (Not Noted As Limitations in the Submission).

Issues for Consideration

• Pembrolizumab is indicated for this patient population; however, it is currently not funded in Canada though it has received regulatory approval. As pembrolizumab is not currently funded, its use is relatively uncommon at present, as noted by drug plans; however, its use is expected to increase quickly over the next few years among patients with private insurance. The cost-effectiveness of dostarlimab compared with pembrolizumab is unknown.
• The current submission is for the use of dostarlimab monotherapy for the treatment of adults with recurrent or advanced dMMR or MSI-H EC that has progressed on or after prior treatment with a platinum-containing regimen. Based on clinical experts’ responses to drug plan feedback, there is the potential for indication creep among patients who received platinum-based therapy only in early-stage disease, those who received more than 2 lines of therapy for advanced or recurrent disease, and those who are currently receiving systemic therapy for recurrent dMMR or MSI-H (with prior platinum-based therapy). Off-label use of dostarlimab among patients with advanced or recurrent EC is likely to result in an increased budget impact.
• Drug plan feedback noted that dMMR and MSI-H testing is not a standard or reflexive test across jurisdictions, The clinical experts consulted by CADTH indicated that all patients with EC of any stage should receive dMMR and MSI-H testing at the time of diagnosis, and that it is not costly and is easily performed.
• The submitted economic model did not include an option to assess a weight-based dosing regimen for dostarlimab; rather, 2 fixed-dosing regimens (i.e., 500 mg every 3 weeks and 1,000 mg every 6 weeks) were considered on the basis of the Health Canada indication and the GARNET study protocol. The clinical experts consulted by CADTH indicated that dostarlimab may be administered as either a flat-rate dose or a weight-based dose, based on the maximum capped dose derived from several pharmacokinetic studies in support of weight-based dosing of immunotherapies. As dostarlimab is an immunotherapy with monoclonal antibodies similar to pembrolizumab, weight-based dosing for dostarlimab is likely to be similarly implemented in Canadian practice, despite the absence of weight-based dosing information in the product monograph and will likely have an impact on associated drug costs. The cost-effectiveness of dostarlimab’s weight-based dosing regimen, however, is unknown.

Overall Conclusions

The CADTH review of the clinical evidence found no direct comparative evidence between dostarlimab and any relevant treatment comparator in the treatment of adults with recurrent or advanced dMMR or MSI-H EC. There is indirect evidence to suggest that dostarlimab may improve PFS, compared with standard of care (chemotherapeutic regimens); however, no strong conclusions can be drawn because of the many limitations in the indirect comparisons. The sponsor’s submitted ITCs were based on data from patients whose dMMR or MSI-H status was unknown, meaning that they are not aligned with the indication for dostarlimab. Thus, the effectiveness of dostarlimab remains highly uncertain relative to any currently reimbursed treatment.

The sponsor submitted a model comparing the cost-effectiveness of dostarlimab with second-line treatments (PLD, doxorubicin, paclitaxel, CMT, carboplatin plus PLD, carboplatin plus paclitaxel, and carboplatin). According to the comparative evidence in the submitted model, carboplatin plus paclitaxel was the next most effective treatment on the efficiency frontier, with the least uncertainty around OS and PFS parameters. The sponsor’s estimate of the pairwise ICER for dostarlimab versus carboplatin plus paclitaxel was $164,193 per QALY, and the probability that dostarlimab is cost-effective was 0% at a willingness-to-pay threshold of $50,000 per QALY.

CADTH was not able to conduct a reanalysis because of foundational limitations with the sponsor’s model and submitted clinical evidence. Notably, the efficacy of comparator treatments in dMMR or MSI-H EC patients was unknown and, thus, the cost-effectiveness of dostarlimab compared with second-line treatments in the indicated population is unknown. The sponsor’s choice of a PSM produced QALY estimates that were likely overestimated. The model was highly sensitive to long-term extrapolation assumptions about OS and PFS, with a sizable majority of incremental QALYs observed in the extrapolated period (i.e., beyond the period of the trial). The sponsor’s model predicts a post-progression survival benefit for dostarlimab that is not supported by the trial evidence. CADTH was unable to address the critical limitation regarding the apparent post-progression survival benefit because of constraints introduced by the submitted model structure. Exploratory analysis found that the model is highly sensitive to several model assumptions, including long-term survival extrapolation, treatment discontinuation, and treatment-effectiveness waning after 24 months. Based on the CADTH exploratory analysis, a price reduction of 83% would be needed for dostarlimab to be considered cost-effective at a willingness-to-pay threshold of $50,000 per QALY; however, this estimate is subject to limitations in the sponsor’s submission, most crucially the inappropriateness of the estimated efficacy of comparator treatments in the dMMR and MSI-H population. Additional price reduction may be warranted.

Treatment of dMMR or MSI-H EC with dostarlimab increases costs compared with other currently available treatments. CADTH could not estimate the amount of health improvement that may result from treatment with dostarlimab because of highly uncertain clinical efficacy data and a lack of information about the efficacy of comparator treatments. Because of the methodological limitations identified in the model and clinical evidence, the cost-effectiveness of dostarlimab is unknown.

Detailed Results of the Sponsor’s Base Case

While the sponsor included multiple comparator treatments to assess the cost-effectiveness of dostarlimab, only the ICER for dostarlimab versus carboplatin + paclitaxel is presented below. The carboplatin + paclitaxel subgroup had the largest number of observations in the RWE survival cohort evidence used within the model. Accordingly, there is the least amount of parametric uncertainty for this comparator. Carboplatin + paclitaxel also had the highest comparative efficacy to dostarlimab (i.e., next-highest QALYs) relative to other subgroups. The true efficacy of carboplatin + paclitaxel within the indicated population remains unknown.

Table 7

Disaggregated Summary of the Sponsor’s Economic Evaluation Results.

Detailed Results of the Sponsor’s Economic Evaluation

Table 8

Summary of the Sponsor’s Economic Evaluation Results (Dostarlimab Versus Pairwise Doxorubicin).

Table 9

Summary of the Sponsor’s Economic Evaluation Results (Dostarlimab Versus PLD).

Table 10

Summary of the Sponsor’s Economic Evaluation Results (Dostarlimab Versus CMT).

Table 11

Summary of the Sponsor’s Economic Evaluation Results (Dostarlimab Versus Carboplatin).

Table 12

Summary of the Sponsor’s Economic Evaluation Results (Dostarlimab Versus Paclitaxel).

Table 13

Summary of the Sponsor’s Economic Evaluation Results (Dostarlimab Versus Carboplatin + PLD).

Scenario Analyses

While CADTH did not conduct any formal reanalyses of the sponsor’s model, the economic review team performed an exploratory analysis to explore the impact of several key limitations on the ICER which included applying an alternate parametric survival distribution for OS; applying an alternate parametric survival distribution for PFS; applying a treatment stopping rule at 2 years with 60% of patients stopping treatment; applying a treatment-waning effect starting at 24 months; and revising the pricing of carboplatin.

Of note, the fundamental limitations in the sponsor’s model persist within this exploratory analysis. There is no direct evidence to support the comparative efficacy of dostarlimab to multiple relevant comparator treatments. Therefore, this exploratory analysis should not be interpreted as a formal CADTH reanalysis to which credence should be given to the results; in particular, the incremental QALY benefit estimated as part of this exploratory analysis remains unlikely to be representative of the true effect of dostarlimab, such that the corresponding ICER is unlikely to be reflective of the true cost-effectiveness of dostarlimab. Instead, the key insight from this exploratory analysis is that the cost-effectiveness estimate of dostarlimab is highly influenced by the uncertainty in the OS and PFS data, and treatment-waning effects applied at 24 months.

Despite the many limitations associated with the sponsor’s submitted ITCs, and that carboplatin + paclitaxel is unlikely to be used in the second-line setting for the indicated population according to the clinical experts consulted by CADTH, carboplatin + paclitaxel was selected as the comparator of choice to explore the cost-effectiveness against dostarlimab in the CADTH exploratory analysis as the number of data points available within the RWE cohort for OS and PFS were greater than other treatments and it demonstrated better efficacy relative to other treatments. However, the cost-effectiveness of dostarlimab compared with all single-agent and combination chemotherapies, and hormone therapy, is highly uncertain. Due to several limitations of the KM data for OS and PFS informed by the available clinical evidence and feedback from the clinical experts consulted by CADTH pertaining to the clinical plausibility of the long-term extrapolations for OS and PFS in the sponsor’s base case, CADTH selected alternate survival distributions for OS and PFS based on clinical plausibility as part of the exploratory analysis. Additionally, CADTH applied a stopping rule at 2 years following treatment initiation with 60% of patients discontinuing treatment. CADTH further applied a treatment-waning effect beginning at year 2, such that patients who initially received dostarlimab would experience similar treatment effects as carboplatin + paclitaxel. When an alternative parametric survival distribution (Exponential) was chosen for the long-term extrapolation of the OS data, the predicted incremental gain in life-years was 82% lower than in the sponsor’s base case, resulting in lower incremental QALYs (0.98) and a higher estimated ICER ($460,362 per QALY). When an alternative parametric survival distribution (Lognormal) was chosen for the long-term extrapolation of the PFS data, the predicted incremental gain in life-years was 82% lower than in the sponsor’s base case, resulting in lower incremental QALYs (0.96) and a higher estimated ICER ($470,639 per QALY).

Table 14

CADTH Revisions to the Submitted Economic Evaluation.

Table 15

Summary of the Stepped Analysis of the CADTH Exploratory Analysis Results.

CADTH undertook several scenario analyses on the CADTH exploratory analysis to determine the impact of alternative assumptions on the exploratory analysis of cost-effectiveness for dostarlimab compared with carboplatin + paclitaxel. This included:

1. Comparison against the most relevant comparators in Canadian clinical practice according to the clinical experts consulted by CADTH: paclitaxel.
2. No treatment waning.
3. Lower health state utility value (0.625) based on the Progressed Disease health state utility value for the ZoptEC trial.¹

The CADTH exploratory analysis was most sensitive to the cost-effectiveness comparison against paclitaxel, the most commonly used treatment in Canadian clinical practice according to CADTH’s clinical experts, which was informed by efficacy data from the sponsor’s indirect treatment comparisons.

Table 16

Summary of CADTH Exploratory Scenario Analyses.

Price-Reduction Analysis

As no formal CADTH reanalysis was performed, price-reduction analyses were conducted using only the sponsor’s base case assumptions. This deterministic analysis was subject to the key limitations of the sponsor’s model as noted in the CADTH Appraisal of the Sponsor’s Economic Evaluation section. Based on the CADTH exploratory analysis, a reduction in the price of dostarlimab by 83% would be required for dostarlimab to be cost-effective at a willingness-to-pay threshold of $50,000 per QALY compared to carboplatin + paclitaxel. It is important to note that this price-reduction estimate is based on estimates of incremental life-years (and hence QALYs) that are highly uncertain and may not be representative of the true incremental effect of dostarlimab. Consequently, the price reduction required for dostarlimab to be cost-effective remains unknown.

Table 17

Price-Reduction Analyses of Sponsor’s Base-Case and CADTH Exploratory Analysis.

Summary of Sponsor’s BIA

The sponsor submitted a budget impact analysis (BIA) estimating the incremental budget impact of reimbursing dostarlimab for patients with mismatch repair deficient (dMMR) or microsatellite instability-high (MSI-H) recurrent or advanced endometrial cancer (EC) that has progressed on or following prior treatment with a platinum-containing regimen. The BIA was undertaken from the perspective of the Canadian public drug plans over a 3-year time horizon, and the sponsor’s pan-Canadian estimates reflect the aggregated results from provincial budgets (excluding Quebec). In the reference scenario, patients were assumed to receive doxorubicin, PLD, carboplatin, carboplatin + paclitaxel, carboplatin + PLD, paclitaxel, and hormonal therapies, in the second-line setting following failure on first-line platinum-based chemotherapy. In the new drug scenario, dostarlimab was assumed to be reimbursed and prescribed as second-line therapy.²⁴

The sponsor estimated the eligible population using an epidemiologic approach was derived via several assumptions and inputs to first estimate the incident population (i.e., newly eligible patients advanced or recurrent EC patients who meet the eligibility criteria for dostarlimab upon its introduction) and the prevalent population (i.e., previously eligible patients who would have met eligibility criteria for treatment with dostarlimab in the years before its introduction).²⁴ The sponsor assumed that a proportion of patients eligible for treatment at the start of each year would initiate treatment in each quarter of the year, to calculate quarterly costs per year.²⁴ This was accomplished by further providing breakdown of the proportion of patients on dostarlimab or other comparator treatment by progression status at each quarter year using PFS and OS curves for each treatment, respectively. The estimated numbers of patients on each treatment in each quarter were then multiplied by corresponding estimates of the quarterly costs of each treatment to ascertain quarterly treatment-specific costs per year.²⁴

In the sponsor’s base case, drug-acquisition costs, dispensing fees, mark-up costs, costs of subsequent therapy, costs of screening for dMMR/MSI-H, and health care resource utilization costs (i.e., pharmacist, nurse, physician, and infusion chair time) were captured. Drug wastage was considered in calculations, with dosing based on Cancer Care Ontario’s regimen monographs. Drug-administration costs were not included.²⁴

Key inputs to the BIA are documented in Table 19.

Table 19

Summary of Key Model Parameters.

Summary of the Sponsor’s BIA Results

The sponsor estimated that the budget impact of reimbursing dostarlimab as second-line treatment for patients with dMMR/MSI-H recurrent or advanced EC would be $17,209,586 in year 1, $24,690,413 in year 2, and $28,464,128 in year 3 for a 3-year total budget impact of $70,364,128.²⁴

CADTH Appraisal of the Sponsor’s BIA

CADTH identified several key limitations to the sponsor’s analysis that have notable implications on the results of the BIA:

• Uncertainty in the estimated population size eligible for dostarlimab: The sponsor undertook an epidemiological approach to estimate the size of the population eligible for dostarlimab. This required deriving inputs from the published literature and applying several assumptions to derive estimates for the incident and prevalent populations in a multi-step approach. The clinical experts consulted by CADTH indicated that although the final estimated population appeared to be largely reasonable, several estimates of the target population derived from the sponsor’s assumptions and inputs may be associated with some uncertainty. First, the clinical expert consulted by CADTH noted that the sponsor’s assumed annual incidence appeared to be slightly higher than expected. CADTH’s clinical expert indicated that according to 2021 statistics reported by a Canadian Cancer report,²⁵ there will be approximately 8,000 new cases with an age-standardized incidence rate of 37.2 per 100,000 women. Additionally, 1 of the experts noted that the sponsor’s assumed proportion of patients with Stage IV progression who would receive active therapy (70%) appeared to be reasonable while the other noted that this proportion may be lower (50%). Additionally, the clinical experts consulted by CADTH indicated that the assumed proportion of patients with dMMR/MSI-H could be as high as 30%.
  • CADTH undertook several scenario analyses to explore the uncertainty in the estimated population size. In scenario analyses, CADTH explored alternate assumptions by (a) decreasing the annual incidence of uterine cancer to 37.2%; (b) decreasing the proportion of patients with Stage IV disease progression requiring active therapy to 50%; and (c) applied the alternate assumption that 30% of patients will have a dMMR or MSH-H status.
• Exclusion of relevant comparators: As per the Health Canada indication and the sponsor’s submitted reimbursement request, the submitted pharmacoeconomic model for dostarlimab is indicated for the treatment of monotherapy for the treatment of adult patients with dMMR/MSI-H recurrent or advanced EC that has progressed on or following prior treatment with a platinum-containing regimen. Feedback from the clinical experts consulted by CADTH for this review indicates that the sponsor excluded several chemotherapy agents from the BIA (monotherapies and combined treatment regimens) that were identified to be relevant comparators for the indicated population (i.e., pembrolizumab, cisplatin, carboplatin plus docetaxel, carboplatin plus doxorubicin, and cisplatin plus doxorubicin). Additionally, the sponsor included hormonal therapies (i.e., medroxyprogesterone, megestrol, tamoxifen, anastrozole, and exemestane), which is not a relevant comparator according to the clinical experts consulted by CADTH and is commonly offered to patients with a more limited recurrence risk (e.g., low volume disease burden with an estrogen/progesterone receptor positive). As such, the incremental budget impact of dostarlimab is associated with uncertainty.
  • CADTH was unable to address this limitation.
• The market share distribution of comparators in the reference scenario is uncertain: In the sponsor’s submitted BIA, the baseline market share distribution for the comparator treatments in the reference scenario did not reflect clinical expectations of treatments offered in the second-line setting, as indicated by CADTH’s clinical expert and drug plan feedback. CADTH’s clinical expert noted that commonly used second-line therapies in the Canadian clinical practice include paclitaxel or liposomal doxorubicin, however, in the sponsor’s base case, carboplatin + paclitaxel was assigned the largest market share while doxorubicin and paclitaxel were assigned the smallest market share. CADTH’s clinical expert further indicated that the purpose for first-line chemotherapy (either for metastatic disease, or for recurrence management) is suggestive of the expected market share distribution of second-line chemotherapy. Specifically, the clinical experts consulted by CADTH noted that patients who progress after first-line therapy do so quickly, in which case a platinum doublet such as carboplatin + paclitaxel would not be used unless patients were platinum-sensitive. Given that the indicated population is dMMR/MSI-H, these patients have aggressive disease and are platinum-refractory rather than platinum-sensitive, thus, it is unlikely for such a high proportion of patients to be retreated with carboplatin + paclitaxel.
  • CADTH did not address this limitation. Due to the uncertainty in the baseline market share distribution in the reference scenario, CADTH revised the baseline market share distribution to reflect the clinical expert opinion in a scenario analysis.
• The anticipated uptake of dostarlimab in the new drug scenario is uncertain: The sponsor assumed that dostarlimab would have an anticipated market uptake rate of 50% in year 1, 70% in year 2, and 70% in year 3 while displacing comparator therapies proportionally to their estimated market shares in the reference scenario (i.e., market share for all other comparators would be halved in year 1, further reduced by 20% in year 2, and remain steady after that). The clinical expert consulted by CADTH indicated that while the sponsor’s anticipated uptake rate of dostarlimab in the new drug scenario appeared to be reasonable, there remains uncertainty to how dostarlimab will displace other comparators, as dostarlimab is expected to likely displace hormonal therapies and doxorubicin more than others given the low response rates of other treatments. Additionally, CADTH’s clinical expert noted that the introduction of dostarlimab would fully displace carboplatin + paclitaxel and
  • CADTH addressed this limitation by revising the market share distribution in the new drug scenario to reflect the expert’s feedback.
• The sponsor assumed continued treatment over the 3-year time horizon, which likely overestimated drug costs: In the base case, the sponsor assumed that all patients would continue to remain on dostarlimab over the 3-year time horizon. The clinical experts consulted by CADTH indicated that approximately 50 to 60% of patients would likely stop treatment after 2 years of treatment. Based on expert feedback, CADTH revised the treatment stopping rule to 24 months, and revised the proportion of patients expected to stop treatment at 2 years to 60%.
  • CADTH addressed this limitation by permitting 60% of patients to stop treatment at 2 years.
• The sponsor’s submitted BIA is based on the same clinical evidence (i.e., OS and PFS) as its pharmacoeconomic analysis, and may overestimate drug costs: The sponsor incorporated extrapolated OS and PFS curves up to 48 months (reflecting baseline and years 1, 2 and 3) to estimate the proportion of patients who remained alive, and those who were on treatment, respectively, in each year over the 3-year time horizon in their submitted BIA. The sponsor applied the same treatment-specific OS and PFS curves as in their pharmacoeconomic base case to calculate quarterly drug costs over the 3-year time horizon in their submitted BIA. As several limitations were identified with the efficacy data in the GARNET study, the long-term extrapolations of this clinical evidence are highly uncertain. As such, CADTH opted to use trial evidence only (i.e., Kaplan-Meier OS and PFS data from the GARNET study), as it represents OS and PFS estimates without any statistical adjustments. CADTH selected the relevant time points from this data to align with the forecasted period in the BIA. OS and PFS estimates from year 2 or 24 months onwards were assumed to remain the same for the remainder of the time horizon in the BIA.
  • CADTH addressed this limitation by applying the unadjusted Kaplan-Meier data from the GARNET study for OS and PFS in the BIA case. In a scenario analysis, CADTH explored the impact of arbitrarily assuming less efficacious estimates for OS by 20% at each time point.

CADTH Reanalyses of the BIA

Table 20

CADTH Revisions to the Submitted Budget Impact Analysis.

Applying the changes in Table 20 resulted in a minor increase in the estimated budget impact under the drug plan perspective to $ over 3 years. The results of the CADTH stepwise reanalyses are presented in summary format in Table 21 and a more detailed breakdown is presented in Table 22.

Table 21

Summary of the CADTH Reanalyses of the BIA.

Table 22

Detailed Breakdown of the CADTH Reanalyses of the BIA.

CADTH conducted the following additional scenario analyses from the drug plan perspective (Scenarios 1 to 5, Table 23):

1. Decreasing the annual incidence of uterine cancer to 37.2% based on 2021 Canadian Cancer statistics.²⁵
2. Applying the alternate assumption that approximately 50% of patients will have Stage IV disease with progression and require active therapy.
3. Applied the alternate assumption that 30% of patients will have a dMMR or MSH-H status.
4. Applied the arbitrary assumption that the overall survival is 20% less efficacious at each time point on the Kaplan-Meier OS curve.
5. Applied an 83% reduction in the price of dostarlimab to align with the point at which the ICER is within the willingness-to-pay threshold of $50,000 per QALY in the CADTH exploratory reanalysis.

The model results were most sensitive to changes in the proportion of patients with a dMMR/MSI-H status.

Table 23

CADTH Scenario Analyses.