1 Recommendations

Endometrial cancer starts in the lining of the uterus and is the most common type of uterine cancer. At diagnosis, primary advanced endometrial cancer refers to stage 3 or 4 disease that has spread beyond the uterus. Recurrent endometrial cancer refers to cancer that is detected either radiologically or histologically, when there has been remission after initial treatment. Mismatch repair (MMR) status, the functionality of the DNA MMR system in tumours, is routinely tested in endometrial cancer. About 75% of people with endometrial cancer have tumours that are MMR proficient (MMRp) or microsatellite stable (MSS). In MMRp, the DNA repair mechanisms are intact and mutations can be corrected. In MSS, the length of microsatellites remains unchanged. People from Black ethnic backgrounds have a higher incidence of the p53‑abnormal (p53abn) subtype of endometrial cancer with MSS or MMRp. This may correlate with TP53-mutated (TP53mut) tumours (see section 3.4). This represents a small proportion of all endometrial cancers, but is often more aggressive and associated with poorer outcomes. The clinical experts explained that endometrial cancer with MSS or MMRp is a molecularly heterogeneous group. They suggested that routinely available molecular testing cannot further identify subgroups of endometrial cancer with MSS or MMRp. They highlighted that the median survival for people with endometrial cancer with MSS or MMRp is usually less than 2 years. The patient experts explained that living with stage 3 or 4 endometrial cancer with MSS or MMRp has a substantial impact on all aspects of life for both the person and their family. This includes debilitating physical symptoms, psychological distress from the uncertainty of disease progression and financial burden. The committee acknowledged that primary advanced or recurrent endometrial cancer with MSS or MMRp can have a negative impact on people with the condition, as well as on their families and carers.

Standard care for primary advanced or recurrent endometrial cancer is platinum-containing chemotherapy, typically a combination of carboplatin and paclitaxel, followed by surveillance scans every 12 weeks. People whose cancer progresses after chemotherapy may be offered immunotherapy, further chemotherapy or, for a very small proportion, maintenance hormone treatment. Pembrolizumab plus lenvatinib is available as an option for people who have had treatment for endometrial cancer (see NICE's technology appraisal guidance on pembrolizumab with lenvatinib for previously treated advanced or recurrent endometrial cancer). The clinical experts explained that, once cancer has progressed after chemotherapy, about 35% to 40% of people are unable to tolerate further treatment at second line, including immunotherapy. They highlighted that the side effects are more often related to lenvatinib than to pembrolizumab. The patient experts agreed and emphasised the high unmet need. They highlighted the limited treatment options for endometrial cancer with MSS or MMRp at this stage, which can leave people feeling frustrated, hopeless and abandoned. The committee acknowledged the high unmet need because of limited first-line treatment options for endometrial cancer with MSS or MMRp. It concluded that people with the condition, and their families, would welcome safe and effective treatments that offer durable responses and are well tolerated.

For this evaluation, the company positioned dostarlimab as an add-on treatment to platinum-containing (also referred to as platinum-based) chemotherapy as a first-line option for primary advanced or recurrent endometrial cancer with MSS or MMRp, when systemic treatment is suitable. The company explained that its target population is narrower than the marketing authorisation. The committee noted that dostarlimab is recommended for a different subpopulation in NICE's technology appraisal guidance on dostarlimab with platinum-based chemotherapy for treating primary advanced or recurrent endometrial cancer with high microsatellite instability or mismatch repair deficiency. The company explained that the only relevant comparator is carboplatin plus paclitaxel, which is standard care in the NHS for primary advanced or recurrent endometrial cancer with MSS or MMRp (see section 3.2). It also highlighted that, if recommended, dostarlimab plus platinum-containing chemotherapy would preclude using pembrolizumab-based regimens at second line for this population. This is because immunotherapies are not offered more than once for this condition in the NHS. The clinical experts confirmed that, for the company's target population, the choice of comparator aligns with NHS practice. The committee agreed with the company's positioning of dostarlimab plus platinum-containing chemotherapy and concluded that the relevant comparator is carboplatin plus paclitaxel.

The key clinical-effectiveness evidence used in the company's submission and economic model came from RUBY‑1, an ongoing phase 3 multinational double-blind randomised trial. It compared 18 weeks of dostarlimab plus carboplatin and paclitaxel (from now, platinum-containing chemotherapy) followed by dostarlimab monotherapy for up to 3 years (from now, the dostarlimab arm) with 18 weeks of placebo plus platinum-containing chemotherapy followed by placebo for up to 3 years (from now, the placebo arm). Randomisation was stratified by MMR and MSS status, prior external pelvic radiotherapy and disease status. The trial included 494 adults (aged 18 years and over) with primary stage 3 or 4, or recurrent endometrial cancer, and an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. Of these, 76% (376 of 494) had endometrial cancer with MSS or MMRp. Mutational data was available for 400 people, of which 88 people had TP53mut tumours. The primary endpoints were overall survival (OS) and progression-free survival (PFS), assessed by the investigator according to Response Evaluation Criteria in Solid Tumors v1.1. In this evaluation, the company presented results from 2 planned interim data cuts, which were used in the economic model:

• IA1 (28 September 2022) for PFS, health-related quality of life and time to treatment discontinuation

• IA2 (22 September 2023) with a median follow up of 37.5 months for OS and adverse events.
  
  The company considers the median follow up for IA1 to be commercial-in-confidence, so this information cannot be reported here. The company highlighted that only 54.8% of data maturity in OS was reached in the population with MSS or MMRp at IA2. So, the OS data at this interim analysis was still immature. The EAG highlighted that no further interim data cuts were planned before RUBY‑1's expected completion in around the third quarter of 2026. The EAG highlighted concerns about the reliability of the OS data from RUBY‑1, particularly the data beyond 30 months, because of heavy censoring. The committee took this into account in its consideration.

The clinical experts noted that people in RUBY‑1 were generally younger than the population likely to have dostarlimab in the NHS. But they thought that the RUBY‑1 population was broadly representative of NHS clinical practice. This was because people in the NHS would typically need to have an ECOG performance status of 0 or 1 to be considered fit enough for triplet therapy (chemotherapy and immunotherapy). About 63.6% (239 of 376) of people with endometrial cancer with MSS or MMRp from RUBY‑1 had a subsequent treatment after disease progression. These included chemotherapy, immunotherapy (such as pembrolizumab plus lenvatinib and pembrolizumab monotherapy), radiation therapy, hormone treatment and bevacizumab. At the IA2 data cut, more people in the placebo arm (72.8%, 134 of 184) had a subsequent treatment compared with in the dostarlimab arm (54.7%, 105 of 192). About 27.1% (102 of 376) of people with endometrial cancer with MSS or MMRp had subsequent immunotherapies. More people in the placebo arm (37%, 68 of 184) had subsequent immunotherapies compared with people in the dostarlimab arm (17.7%, 34 of 192). The EAG highlighted that the subsequent treatments used in RUBY‑1 were not fully generalisable to NHS clinical practice. Specifically, bevacizumab monotherapy is not used for endometrial cancer in the NHS. Also, people whose cancer progresses after first-line immunotherapy would not usually have further immunotherapy in later lines of treatment. The committee recalled its discussion on managing endometrial cancer in the NHS (see section 3.2). It noted that the proportion of people with endometrial cancer with MSS or MMRp in RUBY‑1 who went on to have a second-line treatment was likely higher than what would be seen in the NHS. The committee concluded that there was uncertainty on whether the findings of RUBY‑1 were generalisable to people who would likely have dostarlimab added onto platinum-containing chemotherapy for endometrial cancer with MSS or MMRp in the NHS in particular. This was because of differences in subsequent treatments.

The company presented PFS results from the IA1 and IA2 data cuts. At IA1, the median PFS was 9.9 months in the dostarlimab arm compared with 7.9 months in the placebo arm. The difference was statistically significant (hazard ratio [HR] 0.76, 95% confidence interval [CI] 0.59 to 0.98). But the difference was not statistically significant at the IA2 data cut. The company considers the PFS results at IA2 to be commercial-in-confidence and so they cannot be reported here. The company also presented results for PFS2. This was defined as the time from treatment randomisation to the date of assessment of progression on the first subsequent anticancer treatment after study treatment or death by any cause (whichever is earlier) in people with endometrial cancer with MSS or MMRp at IA2. Median PFS2 was 24.6 months in the dostarlimab arm compared with 15.9 months in the placebo arm (HR 0.74, 95% CI 0.57 to 0.97). The EAG noted that results from this analysis showed a median improvement of 8.7 months in the time to a second progression event for people in the dostarlimab arm compared with people in the placebo arm.

The company did not present OS results at IA1 in its submission. At IA2, the median OS was 34.0 months in the dostarlimab arm compared with 27.0 months in the placebo arm. But the difference was not statistically significant (HR 0.79, 95% CI 0.60 to 1.04).

To estimate the cost effectiveness of adding dostarlimab to platinum-containing chemotherapy to treat primary advanced or recurrent endometrial cancer with MSS or MMRp, the company used a partitioned survival model. This had 3 health states (progression-free, progressed disease and death), a 1‑week cycle with no half-cycle correction and a 36‑year time horizon. Data from the placebo arm of RUBY‑1 was used to inform the comparator arm in the model (that is, platinum-containing chemotherapy). In line with the marketing authorisation, a 3‑year stopping rule was applied for dostarlimab (see section 2.2). The committee concluded that the company's model was suitable for decision making.

The company extrapolated the long-term effects of dostarlimab plus platinum-containing chemotherapy, and of platinum-containing chemotherapy alone, on PFS and OS in people with primary advanced or recurrent endometrial cancer with MSS or MMRp beyond the trial data. It assumed non-proportional hazards for PFS and OS. This was because dostarlimab has a different mechanism of action compared with platinum-containing chemotherapy. There was a longer time on treatment with dostarlimab but only 6 cycles of platinum-containing chemotherapy in RUBY‑1, and the commonly observed delayed response with immunotherapies. To extrapolate the long-term effects, it fitted independent parametric distributions to model PFS and OS in the 2 treatment arms. The EAG agreed with the company that the proportional hazard assumption did not hold. It also agreed that fitting independent parametric distributions to model PFS and OS in the 2 arms was reasonable.

To model subsequent treatments, the company adjusted data from RUBY‑1 on subsequent treatment use. This was based on feedback from UK healthcare professionals about the options available in the relapsed setting for endometrial cancer with MSS or MMRp. At the clarification stage, the EAG highlighted that bevacizumab monotherapy is not used in the NHS to treat endometrial cancer. So, the company redistributed the proportions having bevacizumab monotherapy to other subsequent treatment options and, for the platinum-containing chemotherapy arm only, also to immunotherapy (51.2% to pembrolizumab plus lenvatinib). The EAG thought the company's approach of using adjusted RUBY-1 data to reflect NHS practice to be reasonable. It also agreed with the redistribution of bevacizumab monotherapy use in the dostarlimab arm. But it disagreed with the company's redistribution of bevacizumab monotherapy use to immunotherapy in the platinum-containing chemotherapy arm. The EAG explained that bevacizumab monotherapy is not a very effective treatment for endometrial cancer. So, reallocating its use to immunotherapy increased costs without a corresponding clinical benefit. Instead, it preferred to redistribute bevacizumab monotherapy use across all other non-immunotherapy treatments. In the EAG's base case, the proportion of people in the platinum-containing chemotherapy arm having subsequent pembrolizumab plus lenvatinib was 48.8%. It also noted that this redistribution of subsequent treatment had a small impact on the cost-effectiveness estimate.

The committee noted that the unadjusted subsequent immunotherapy use in RUBY‑1 was different between the dostarlimab arm (17.7%, 34 of 192) and the placebo arm (37.0%, 68 of 184), despite the trial's randomised and double-blind design (see section 3.6). The company explained that the dostarlimab arm had fewer progression events, and so lower use of subsequent anticancer treatments. It highlighted that further analysis using data from IA2 showed that 51.2% of people in the platinum-containing chemotherapy arm had subsequent pembrolizumab plus lenvatinib. The company also explained that most progression events occurred within the first 12 months, with some occurring between 12 and 24 months. So, it suggested that the follow-up data from RUBY‑1 may have captured the benefits of subsequent treatment, as shown in the PFS IA2 data.

The clinical experts had differing views on whether the company's or the EAG's subsequent treatment distributions best reflected NHS clinical practice. One clinical expert thought that the company's approach of redistributing bevacizumab monotherapy use to pembrolizumab plus lenvatinib was reasonable. They also noted that bevacizumab and lenvatinib have similar mechanisms of action. The committee recalled its discussion that the subsequent treatment used in RUBY‑1 did not reflect NHS practice (see section 3.5). It was aware that more people in the placebo arm had subsequent immunotherapies in RUBY‑1 compared with people in the dostarlimab arm. But it noted that people in the dostarlimab arm also had second-line immunotherapies which is unlikely to happen in the NHS, and that the company did not adjust the OS hazard ratios for second-line treatments. This meant that the impact of subsequent treatment, particularly the impact of second-line immunotherapies on OS, was unclear. It was also unclear whether the estimates for treatment effect on OS from RUBY‑1 was over- or under-estimated. This could also have had an impact on the assumptions about treatment waning in the longer term (see section 3.12).

Given the uncertainties in the treatment effect seen in RUBY-1, the committee would have preferred to see analyses for OS adjusting for benefits and costs of subsequent treatments, particularly subsequent immunotherapies. It thought that further understanding of the impact of these subsequent treatments on both PFS and OS Kaplan–Meier and extrapolated curves in the dostarlimab and placebo arms, using data from the latest data cut at IA2, would be helpful. It requested that the company provide analyses exploring the impact of subsequent immunotherapies on dostarlimab's treatment effect on both PFS and OS, using data from IA2. For example, this could include:

• using PFS2 (see section 3.6) to address pre- and post-progression survival

• presenting Kaplan–Meier curves for PFS and OS with and without subsequent immunotherapies to assess the potential impact of subsequent immunotherapies on the treatment effect

• presenting Kaplan–Meier curves for PFS and OS that account for the potential differences in the timing of subsequent immunotherapy started across the 2 treatment arms (for example, using treatment switching methods) because people in the 2 arms may have started subsequent immunotherapies at different time points in RUBY‑1, which may have an impact on the treatment effect of dostarlimab.

In its base case, the company assumed that treatment waning was captured in the modelled OS. This assumption was based on RUBY‑1, which the company suggested had shown a sustained OS benefit in the dostarlimab arm compared with the placebo arm. The company also highlighted that its independent modelling of OS curves should have implicitly captured any waning of the treatment effect. The EAG thought that the company's modelling of OS was generally appropriate, and agreed with the company that treatment waning was likely captured in the OS extrapolations. This was because the PFS plots showed that the risk of progression was similar for both arms after about 2 years. But the EAG noted that, at IA2, the Kaplan–Meier curve for OS in the 2 treatment arms appeared to converge from month 30 then diverge again from month 36 onwards. So, it thought that a scenario of gradual treatment-effect waning may be informative. The company also provided 2 treatment-effect waning scenarios: one with a 2‑year stopping rule and another with a 3‑year stopping rule. This was in line with dostarlimab's marketing authorisation (see section 2.2). The committee did not think that treatment-effect waning was implicitly captured in the model. This was because the company's extrapolated OS curves with and without incorporated treatment waning were different. The committee recalled the uncertainties associated with the evidence on OS. These included data immaturity, heavy censoring, the potential impact of subsequent treatments on OS and the lack of adjustment of subsequent treatments on OS (see section 3.11). It thought that there was high uncertainty related to the treatment-effect waning assumptions. It concluded that it would like to have seen evidence and analyses in which:

• including or excluding treatment-effect waning in the model was sufficiently justified

• the impact of second-line treatment effect on OS was adjusted for

• the potential interplay between the impact of subsequent treatments on OS and treatment-effect waning is explored.

To model time on treatment in its base case, the company used weighted completion rates for platinum-containing chemotherapy from RUBY‑1 across both treatment arms during the first 6 cycles. For dostarlimab, it used completion rates for the first 6 cycles, followed by time to treatment discontinuation Kaplan–Meier data adjusted for relative dose intensity up to 3 years. The EAG thought that using completion rates did not fully capture the cost of starting treatment in either arm. This was because the intention-to-treat population included people who were randomised but did not start treatment. So, the completion rate for the first treatment cycle in the model was less than 100%. To capture the full treatment costs in its base case, the EAG preferred to use time to treatment discontinuation Kaplan–Meier data for both arms. In addition, for dostarlimab, it used the relative dose intensity from cycle 1 up to 3 years. The committee concluded that the EAG's approach to modelling time on treatment was appropriate.

In its base case, the company modelled health-state resource use for the dostarlimab arm based on advice from 6 UK healthcare professionals. It assumed that, after the first 18 weeks of treatment, people having dostarlimab who were progression-free would have specific resource use that differed from that in the platinum-containing chemotherapy arm. The EAG disagreed and thought that, after 3 years of dostarlimab monotherapy, people who were progression-free would incur the same resource use as that in the platinum-containing chemotherapy arm after the first 18 weeks of treatment. The clinical experts thought that the level of surveillance would be similar for people who are progression-free after treatment, regardless of whether they have had dostarlimab or platinum-containing chemotherapy. At the committee meeting, the company explained that it agreed with the EAG's modelling. The committee concluded that it preferred the EAG's approach to modelling health-state resource use for the dostarlimab arm.

In its base case, the company included an oral administration cost for lenvatinib to account for specialist oversight related to procurement, prescribing, dispensing and administration. The EAG disagreed and excluded this cost from its base case. It noted that, based on published advice, people are likely to take lenvatinib at home, so there would likely be no administration cost to the NHS. It highlighted that these administration costs only affect the platinum-containing chemotherapy arm because pembrolizumab plus lenvatinib is not used after dostarlimab treatment. The clinical experts agreed that there is likely no cost associated with administering lenvatinib. But they noted that there would be costs related to managing its side effects. The EAG confirmed that these were accounted for in the monitoring costs already included in the model. The committee concluded that the oral administration cost of lenvatinib should be excluded from the model.

The committee's preferred assumptions were to:

• use the EAG's approach to model time on treatment (see section 3.13)

• use the EAG's approach to model health-state resource use for the dostarlimab arm (see section 3.14)

• exclude the cost of oral administration for lenvatinib (see section 3.15).

NICE's manual on health technology evaluations notes that, above a most plausible incremental cost-effectiveness ratio (ICER) of £20,000 per quality-adjusted life year (QALY) gained, judgements about the acceptability of a technology as an effective use of NHS resources will take into account the degree of certainty around the ICER. The committee will be more cautious about recommending a technology if it is less certain about the ICERs presented. But it will also take into account other aspects including uncaptured health benefits. The committee noted the high level of uncertainty in the evidence and company's modelling, specifically that:

• longer data cut of PFS was not statistically significant and OS data was immature (see section 3.6)

• subsequent treatments from RUBY‑1 may not be generalisable to NHS clinical practice and its impact on OS estimates (see section 3.5 and section 3.11)

• the extrapolation of PFS (see section 3.9).

• the extrapolation of OS (see section 3.10).

• modelling of subsequent treatment (see section 3.11).

• whether treatment-effect waning has been implicitly included in the model (see section 3.12).
  
  The committee thought that the further analyses requested (see section 3.19) would likely affect these uncertainties. So, the committee concluded that it did not have an acceptable ICER level.

The committee decided that there were several areas of uncertainty (see section 3.18). It would like clarification, further evidence and analyses:

• using the more mature and most recent IA2 data cut to extrapolate PFS (see section 3.9)

• for OS adjusting for benefits and costs of the second-line treatments that reflect NHS clinical practice and using PFS data at IA2 (see section 3.10 and section 3.11)

• in which including or excluding treatment-effect waning in the model is sufficiently justified, and the potential interplay between the impact of subsequent treatments on OS and treatment-effect waning is explored (see section 3.12)

• specific for the p53‑abnormal or TP53mut subgroups, including cost-effectiveness estimates, that include relevant diagnostic testing costs (see section 3.4, section 3.6 and section 3.20).

The committee noted that there was little difference between the cost-effectiveness estimates in the company's and the EAG's base cases. The exact figures cannot be reported because of confidential discounts for dostarlimab, pembrolizumab and lenvatinib. But both the company's and the EAG's base case ICERs were higher than what NICE normally considers an acceptable use of NHS resources. The committee thought that there were important uncertainties that need to be addressed to inform decision making (see section 3.18 and section 3.19). So, it was unable to conclude that dostarlimab was a cost-effective option for routine commissioning to treat primary advanced or recurrent endometrial cancer with MSS or MMRp in adults when systemic treatment is suitable.

Stakeholders highlighted that people from Black ethnic background have a higher incidence of the more aggressive p53‑abn subtype of endometrial cancer with MSS or MMRp, which is associated with poorer outcomes (see section 3.1). The committee requested further analyses for consideration in the p53‑abnormal or TP53mut subgroups (see section 3.19).

The committee acknowledged that there is a high unmet need for early treatment for people with primary advanced or recurrent endometrial cancer with MSS or MMRp (see section 3.2). It acknowledged that this would be the first immunotherapy available for this condition at first line. It considered whether there were any uncaptured benefits of dostarlimab plus platinum-containing chemotherapy. It did not identify additional benefits of dostarlimab plus platinum-containing chemotherapy not captured in the economic modelling. So, the committee concluded that all additional benefits of dostarlimab plus platinum-containing chemotherapy had already been taken into account.

The committee noted that the company's and the EAG's base case ICERs were higher than what NICE normally considers an acceptable use of NHS resources. It recalled that there were important uncertainties that need to be addressed to inform decision making. So, dostarlimab plus platinum-containing chemotherapy could not be recommended to treat primary advanced or recurrent endometrial cancer with MSS or MMRp in adults when systemic treatment is suitable. Also, further evidence and analyses are needed.