Mirvetuximab soravtansine for treating folate receptor-alpha-positive platinum-resistant epithelial ovarian, fallopian tube or primary peritoneal cancer

There are no known specific causes of epithelial ovarian, fallopian tube or primary peritoneal cancer (from here on referred to collectively as ovarian cancer). Risk increases with age, a family history of ovarian cancer and inherited gene mutations. Early-stage ovarian cancer is underdiagnosed because it usually causes vague or no symptoms. This means that most people have advanced cancer at diagnosis. Symptoms at diagnosis include gastrointestinal issues, ascites, pleural effusion and venous thromboembolism. Symptoms worsen as the condition progresses. While the condition often responds to initial treatment, most people's cancer goes on to develop resistance to chemotherapy. Platinum resistance is defined as disease progression within 6 months after a platinum-based chemotherapy regimen. This can occur after 1 or more courses of platinum-based chemotherapy. The clinical experts explained that there is a lack of effective treatment options for people with platinum-resistant ovarian cancer. The committee concluded that there is a need for new treatments for people with platinum-resistant ovarian cancer.

Ovarian cancer has a substantial impact on quality of life for people with the condition and their families. Statements submitted by the patient experts explained that many people who are diagnosed are in their 50s and 60s, and lead active lives that include work and family responsibilities. Physical symptoms caused by the condition include ascites and gastrointestinal issues. A patient expert at the first committee meeting explained that chemotherapy can have long-lasting side effects and that these had a severe negative impact on their quality of life. They said that chemotherapy made them feel very low and tired. They added that they lost their sense of taste and smell, and had a poor appetite as a result. The patient experts also said that knowing there is a lack of effective treatment options for platinum-resistant ovarian cancer can have a severe psychological and emotional impact. They said that a new treatment option that could improve survival and quality of life in a population with limited options would offer hope.

During the draft guidance consultation period, patient groups collected data using surveys on the impact of the condition, its treatment and the availability of new treatment options. Survey respondents emphasised the poor prognosis after developing platinum-resistant ovarian cancer and the limited treatment options at this stage of the disease. They explained that the side effects of chemotherapy include extreme fatigue, peripheral neuropathy, weakness, breathlessness, nausea, pain, extreme anxiety and hair loss. They added that side effects are debilitating and impact heavily on all aspects of life and that some of these (particularly fatigue and neuropathy) can continue long after chemotherapy treatment has finished. Survey respondents having chemotherapy described having their life on hold, living in isolation, no longer seeing friends or doing hobbies, being unable to do housework, giving up jobs, and relying heavily on others for support. At the second committee meeting, a patient expert emphasised the substantial quality of life impact of having repeated cycles of chemotherapy. They agreed that side effects of chemotherapy are debilitating and life encompassing, and impact both people with the condition and their friends and family. They added that any treatment that would allow people to take a break from chemotherapy is desperately needed. Patient group survey respondents explained that for the few people who had had mirvetuximab soravtansine (from here on referred to as mirvetuximab), most reported having a substantially improved quality of life with fewer side effects. People were more able to return to normal activities, contribute to their community and continue to work, allowing their support system to do the same. The committee recognised the substantial work done by the patient groups in conducting surveys, which provided useful information on the burden of the disease, the impact of chemotherapy treatment and the potential benefits of mirvetuximab. The committee concluded that there is a high disease burden for people with platinum-resistant ovarian cancer, and the availability of a new treatment would benefit people with the condition.

NICE has recommended pegylated liposomal doxorubicin (PLD) and paclitaxel for treating recurrent ovarian cancer (see NICE's technology appraisal guidance on topotecan, pegylated liposomal doxorubicin hydrochloride, paclitaxel, trabectedin and gemcitabine for treating recurrent ovarian cancer, from here on referred to as TA389). The company used pooled chemotherapy (originally, PLD, paclitaxel and topotecan) as the comparator in its analysis. It used data from MIRASOL for the proportions of people having each individual chemotherapy treatment. In MIRASOL, 40.7% of people had paclitaxel, 35.8% had PLD and 23.5% had topotecan. The EAG noted that topotecan is not recommended within its marketing authorisation for treating recurrent platinum-resistant or platinum-refractory ovarian cancer in TA389. So it assumed that the proportion of people having topotecan would be re-distributed over the remaining treatments so that 46.8% people would have PLD and 53.2% would have paclitaxel. The clinical experts said that topotecan is rarely used in clinical practice. In response to draft guidance consultation, the company updated its model in line with the EAG's approach. The committee concluded that the relevant comparator was pooled chemotherapy, including PLD and paclitaxel.

The company presented clinical evidence for mirvetuximab from 3 clinical trials: MIRASOL, FORWARD‑1 and SORAYA. MIRASOL and FORWARD‑1 were phase 3, randomised, open-label studies, and SORAYA was a phase 2, single-arm study. The comparator in MIRASOL and FORWARD‑1 was investigator's choice of chemotherapy, including paclitaxel, PLD or topotecan (from here on referred to as pooled chemotherapy). The key clinical evidence for mirvetuximab came from MIRASOL (n=453). This was because the population in FORWARD‑1 included a broader range of folate receptor (FR)‑alpha expression than specified in the marketing authorisation for mirvetuximab. Also, SORAYA did not have a comparator arm. The primary outcome of MIRASOL was investigator-assessed progression-free survival (PFS). Other key outcomes included overall survival (OS), objective response rate and EQ‑5D‑5L.

The EAG highlighted uncertainties with the generalisability of MIRASOL to UK clinical practice. It noted that the median age of people in MIRASOL was lower than expected in NHS clinical practice. The median age at baseline in MIRASOL was 64 years in the mirvetuximab arm and 62 years in the pooled chemotherapy arm. The EAG highlighted that in Pickwell-Smith et al. (2025), a retrospective cohort study of people diagnosed with ovarian cancer in England between 2016 and 2017, the median age at diagnosis was 68 years. At the first committee meeting, the NHS England Cancer Drugs Fund clinical lead (from here on referred to as the Cancer Drugs Fund lead) said that, over the previous 12 months, the average age of people starting first-line maintenance treatment with poly-ADP polymerase (PARP) inhibitors for ovarian cancer in England was 69 years. The clinical experts said that, in clinical practice, the mean age at diagnosis of people with ovarian cancer was 66 years. This was based on the National Ovarian Cancer Audit State of the Nation Report published in 2025. They said that age of diagnosis is unlikely to affect outcomes in ovarian cancer and that it is common for people included in clinical trials to be younger than in clinical practice. At the second committee meeting, the clinical experts also felt that the age of people diagnosed with ovarian cancer was reducing over time. During the consultation period, SACT data became available (see section 3.8). This showed that the mean age of people starting PLD or paclitaxel treatment in NHS clinical practice was 67 years.

The EAG's clinical advisers noted that the proportion of people who had had treatment with a PARP inhibitor was likely to be lower in MIRASOL than in clinical practice. In MIRASOL, 54.6% of people in the mirvetuximab arm and 57.0% of people in the pooled chemotherapy arm had had a PARP inhibitor. The EAG explained that the results of MIRASOL could have overestimated the benefits of chemotherapy. This was because prior treatment with PARP inhibitors can reduce sensitivity to subsequent chemotherapy after disease progression. The clinical experts thought that the proportion of people previously treated with a PARP inhibitor in MIRASOL was representative of clinical practice. They also said that they did not expect prior PARP inhibitor use to affect outcomes.

The EAG also highlighted that 23.5% of people in the comparator arm of MIRASOL had topotecan, but clinical advice indicated that topotecan is rarely used in clinical practice. The EAG said that topotecan has similar effectiveness to other chemotherapy regimens, so this was not expected to affect the generalisability of MIRASOL. The clinical experts confirmed that topotecan has similar clinical effectiveness to PLD and paclitaxel. The committee noted that there may be some differences between the MIRASOL population and the expected population in NHS clinical practice. But it concluded that these were unlikely to affect the generalisability of MIRASOL. So, it concluded that MIRASOL was acceptable for decision making.

In MIRASOL, mirvetuximab showed a statistically significant improvement in PFS and OS compared with pooled chemotherapy. The hazard ratio (HR) for PFS was 0.63 (95% confidence interval [CI] 0.51 to 0.79) and for OS was 0.68 (95% CI 0.54 to 0.84). The company also provided a post-hoc analysis of FORWARD‑1 that included only people who met the marketing authorisation requirement for FR‑alpha expression. The HR for PFS in the FORWARD‑1 post-hoc analysis was 0.65 (95% CI 0.41 to 1.03) and for OS was 0.66 (95% CI 0.40 to 1.08). It also provided a meta-analysis that pooled the results of MIRASOL and the FORWARD‑1 post-hoc analysis. The EAG noted that the results of the meta-analysis were statistically significant and consistent with the results of MIRASOL and FORWARD‑1. The committee concluded that mirvetuximab improved OS and PFS compared with pooled chemotherapy for treating FR‑alpha-positive platinum-resistant ovarian cancer.

The company said that the subgroup analyses in MIRASOL were exploratory only and were not appropriate for decision making. But the EAG noted that there may be clinical rationale for considering subgroups based on length of primary platinum-free interval. The primary platinum-free interval is the period after the first course of platinum-based chemotherapy until disease recurrence or progression. A primary platinum-free interval of 6 months or less suggests that the ovarian cancer had responded poorly to first-line treatment. The EAG's clinical experts said it was plausible that a longer primary platinum-free interval would lead to a better response to mirvetuximab treatment. In MIRASOL, the OS HR for the subgroup of people with a primary platinum-free interval of more than 6 months was 0.54 (95% CI 0.42 to 0.71). It was 1.07 (95% CI 0.74 to 1.56) for the subgroup of people with a primary platinum-free interval of 6 months or less. The EAG said that the subgroup analyses should be interpreted with caution because the PFS results were not consistent with the OS results. The PFS HR for the subgroup of people with a primary platinum-free interval of more than 6 months was 0.62 (95% CI 0.48 to 0.80). It was 0.68 (95% CI 0.47 to 0.98) for the subgroup of people with a primary platinum-free interval of 6 months or less. The committee noted that the OS HR for the subgroup with a primary platinum-free interval of more than 6 months was more favourable for mirvetuximab than the OS HR for the whole MIRASOL population. The committee acknowledged there was uncertainty associated with the subgroup results and concluded that it would prefer to consider mirvetuximab for the whole population in MIRASOL.

During the draft guidance consultation period, NICE and the National Disease and Registration Service produced a report based on SACT data. The aim of this report was to provide further evidence on the average age and OS for people having standard treatment for platinum-resistant ovarian cancer in the NHS. The SACT report included data from 7,083 adults with epithelial ovarian, fallopian tube or primary peritoneal cancer having PLD or paclitaxel monotherapies in England. Given the positioning of PLD and paclitaxel in the treatment pathway, the report authors thought that this represented a population with advanced platinum-resistant disease. The maximum duration of follow up for this data was 113 months, and median OS was 8.54 months. One clinical expert at the second committee meeting noted that the MIRASOL trial only included people with FR‑alpha-positive tumours, whereas the SACT data was for a population where FR‑alpha status was unspecified. They added that the prognostic impact of FR‑alpha status was unclear and that this may limit generalisability of SACT to the population of interest. But they noted that outcomes were not expected to be much better for the FR‑alpha-positive group. The committee welcomed the use of SACT data and noted that this was provided for a large sample of people having standard treatments in NHS practice. The committee noted the clinical expert's comments that the population in SACT was not solely FR‑alpha-positive. But it concluded that the SACT data was broadly generalisable to outcomes in NHS practice and could be used for decision making.

The company presented a partitioned survival model with 3 health states: pre-progression, post-progression and death. The model had a 1-week cycle length and included a half-cycle correction for PFS and OS. The EAG thought the model structure was appropriate given the high level of maturity of the observed data. The committee concluded that the model structure was appropriate.

The company presented evidence from the September 2024 data cut of MIRASOL, with a median OS follow up of 30.49 months. At the end of follow up, 44 people in the mirvetuximab arm and 18 people in the chemotherapy arm were alive. To extrapolate beyond the end of the trial follow-up period, the company fitted independent parametric models to the mirvetuximab and pooled chemotherapy OS data from MIRASOL. In its base case, the company chose the log-logistic distribution for mirvetuximab OS and the Weibull distribution for pooled chemotherapy OS. The company consulted clinical experts, who expected survival rate in the mirvetuximab arm to be between 8% and 12% at 5 years. But they were uncertain about the survival rate at 10 years. They expected survival in the pooled chemotherapy arm to be between 0% and 5% at 5 years and 0% at 10 years. For mirvetuximab, the company said that the log-logistic distribution implied a hazard function that was consistent with the observed hazard in the mirvetuximab arm (initially increasing then decreasing). The log-logistic distribution predicted a 5-year survival rate of 10%, which the company noted was in the range that its clinical experts had predicted. The 10-year survival rate predicted using the log-logistic distribution was 3%, which the company said was plausible given mirvetuximab's mechanism of action. For pooled chemotherapy, the company said that the Weibull distribution had the best statistical fit. It was also considered plausible by its clinical experts because it aligned with their prediction of 0% to 5% survival at 5 years. The EAG initially preferred to use the gamma distribution for both mirvetuximab and pooled chemotherapy OS in its base case. This was because it provided a good statistical and visual fit, and that the company's clinical experts considered it to be plausible. During the first committee meeting, the clinical experts said that survival for people having chemotherapy is very poor. They said that it was difficult to predict the long-term survival of people having mirvetuximab, but it was plausible that about 10% could live beyond 5 years. This was because of mirvetuximab's novel mechanism of action. The committee noted the uncertainty associated with the long-term survival of people having mirvetuximab. It also thought that it would be useful to have alternative data sources for pooled chemotherapy to help validate the pooled chemotherapy OS extrapolations.

At consultation, the EAG updated its preferred approach to OS extrapolation in light of the SACT data (see section 3.8). It noted that the SACT data provided estimates of survival for people having pooled chemotherapy that were generalisable to NHS clinical practice. It added that the SACT data came from a large sample size with long-term follow up. It noted that a log-logistic distribution provided the best statistical fit and a good visual fit to SACT data for chemotherapy. This indicated that long-term curves that allowed for a decrease in hazards were likely to be most appropriate. The EAG noted that the log-logistic curve applied to the MIRASOL OS data for pooled chemotherapy provided the closest fit to the survival trends in the observed chemotherapy SACT data in the long term. The EAG preferred the log-logistic curve over the company's original Weibull or EAG's original gamma curve. So, the EAG used a log-logistic distribution fitted to MIRASOL OS data for the pooled chemotherapy arm. It also used the log-logistic distribution for the mirvetuximab arm (as per the company base case). This maintained the principle of assuming the same long-term survival profile for both treatments unless there is good reason to assume otherwise. The EAG also provided a scenario that used SACT data directly. This scenario used a log-logistic distribution fit to SACT data directly to model OS for pooled chemotherapy, with the MIRASOL OS HR of 0.68 applied to model OS for mirvetuximab (SACT plus HR approach). The company maintained its original base-case curves fit to MIRASOL data, but agreed that SACT data adequately captured the survival over time for people having chemotherapy. It considered that the EAG's SACT plus HR scenario was an informative scenario for including the SACT data. The EAG noted that the SACT plus HR approach was valid. But it explained that it had not applied this approach in its base case because there were some limitations with it. It was not clear whether the proportional-hazards assumption held over time and there was no SACT data available to model PFS. The EAG also noted that the post-progression survival gain in both the company base case using MIRASOL, and the SACT plus HR scenario, may not be captured appropriately. This was because it was substantially greater than pre-progression survival gain.

At the second committee meeting, the clinical experts explained that the post-progression survival gain for mirvetuximab when applying the SACT plus HR approach was plausible because the benefit of mirvetuximab could continue after progression. The committee agreed that SACT was a large, robust dataset that provided OS outcomes that were representative of NHS practice for people having pooled chemotherapy. It also noted the views of the clinical experts around the post-progression survival gain for mirvetuximab. The committee added that applying the HR from MIRASOL to SACT data to estimate OS for mirvetuximab was appropriate because there was no strong indication that the proportional-hazards assumption did not hold. But, it noted that there were still some uncertainties with this approach, including with:

• the proportional-hazards assumption, and whether this holds over time

• the approach of applying a trial derived HR to SACT data

• the approach of applying a Cox proportional-hazards model to a log-logistic curve, and

• the lack of PFS data from SACT for this evaluation.
  
  The committee concluded that it preferred the SACT plus HR approach for modelling OS. This used the log-logistic distribution fit to SACT data for pooled chemotherapy with the MIRASOL OS HR of 0.68 applied for mirvetuximab (SACT plus HR approach). But the committee noted that there was still uncertainty with using this approach.

The company included treatment-dependent utility values for the pre- and post-progression health states. At the first committee meeting, it preferred to use health-state utility values from MIRASOL in the mirvetuximab arm and from Havrilesky et al. (2009) in the pooled chemotherapy arm. The company said that the utility values from MIRASOL may have overestimated the health-related quality of life (HRQoL) of people having chemotherapy. This was because the EQ‑5D questionnaire was only administered at the start of each chemotherapy cycle. So, it may not have captured the impact of the side effects of chemotherapy. In the pre-progression state, the utility value from MIRASOL was 0.737 for mirvetuximab and 0.706 for pooled chemotherapy. The company preferred to use a pre-progression health-state utility value for pooled chemotherapy of 0.500. In the post-progression state, the utility values from MIRASOL were 0.655 for mirvetuximab and 0.625 for pooled chemotherapy. The company preferred to use a post-progression health-state utility value for pooled chemotherapy of 0.400.

The EAG said that the company's approach was not consistent with the NICE reference case. This was because the company's preferred utilities for pooled chemotherapy at the first committee meeting from Havrilesky et al. were based on a time trade-off exercise using vignettes. The EAG noted that section 4.3.4 in NICE's manual on technology appraisal and highly specialised technologies guidance states that the valuation of HRQoL measured by people with the condition should be based on a valuation of public preferences from a representative sample of the UK population. The EAG also noted some inconsistencies in the results from Havrilesky et al. It also thought that the company's approach was implausible and lacked face validity. The EAG thought that the timing of the EQ‑5D measurement in MIRASOL was not a meaningful source of bias. It compared the utilities from MIRASOL with those from the OVA-301 clinical trial (considered in TA389), in which the EQ‑5D was measured at the start and the end of each treatment cycle. The pooled chemotherapy utilities in OVA-301 (0.718 in the pre-progression state and 0.649 in the post-progression state) were similar to those in MIRASOL (0.706 in the pre-progression state and 0.625 in the post-progression state). The EAG concluded that the health-state utility values from MIRASOL showed a difference in HRQoL between the mirvetuximab and chemotherapy arms that was more likely to be plausible than the values in the company's approach. So, the EAG preferred to use the utility values from MIRASOL in its base case.

The patient experts at both the first and second committee meeting emphasised that people having mirvetuximab have a substantially better quality of life compared with people having chemotherapy. One patient expert said that, when they had chemotherapy, they felt tired and low because of the side effects and had to take time off work. But they said that with mirvetuximab they were able to live a relatively normal life and felt significantly less anxious and depressed. The patient experts also said that people have fewer and less-serious side effects with mirvetuximab than with chemotherapy. Respondents to the surveys conducted by patient groups agreed. They explained that people having chemotherapy for platinum-resistant ovarian cancer have a high symptom burden as well as substantial toxicity from chemotherapy treatments. They added that mirvetuximab improves response rates, symptom burden and toxicity allowing people to live more fulfilled, better quality lives than with chemotherapy. Clinical experts at the first committee meeting noted that any ocular side effects of mirvetuximab tended to be resolved quickly. They highlighted that the proportion of people who stopped treatment because of these was small.

The committee agreed with the EAG that the company's preferred source for the utility values in the pooled chemotherapy arm was not consistent with the NICE reference case. It also agreed that the company's approach lacked face validity because of the large difference in the mirvetuximab post-progression utility and the pooled chemotherapy pre-progression utility. The committee highlighted that the MIRASOL utility values were consistent with the utility values used in previous NICE technology appraisals. The EAG provided a scenario analysis that used the pre- and post-progression utility values from TA389 for both treatment arms. The committee noted that the incremental cost-effectiveness ratio (ICER) for this scenario was higher than in the EAG base case, which used the MIRASOL utility values. It noted that the MIRASOL utility values suggested a small difference in HRQoL between people having mirvetuximab and people having chemotherapy. But, the committee noted that the difference in HRQoL due to differences in side effects between treatments was accounted for when disutilities associated with adverse events were included, as in the EAG's base case. It thought that the MIRASOL utilities were more methodologically robust than the utilities from Havrilesky et al. for pooled chemotherapy preferred by the company. So, at the first committee meeting, the committee concluded that it was more appropriate to use the MIRASOL health-state utility values in both treatment arms. It also preferred to include the disutilities associated with adverse events to capture the impact of side effects on HRQoL.

In response to the draft guidance consultation, the company accepted the limitations of using Havrilesky et al. utilities for pooled chemotherapy. But it maintained that the utility difference between mirvetuximab and pooled chemotherapy in MIRASOL was not clinically plausible. The company updated the utilities in its model based on results of an updated utility regression analysis of MIRASOL data, which included an interaction term between treatment and progression. In the pre-progression state, the utility value from MIRASOL was 0.732 for mirvetuximab and 0.712 for pooled chemotherapy (compared with 0.737 for mirvetuximab and 0.706 for pooled chemotherapy, without the interaction term). In the post-progression state, the utility values from MIRASOL were 0.675 for mirvetuximab and 0.596 for pooled chemotherapy (compared with 0.655 for mirvetuximab and 0.625 for pooled chemotherapy, without the interaction term). The company stated that the results of this analysis supported the clinical understanding that mirvetuximab provides lasting benefits over chemotherapy even as the disease progresses. The company also provided a scenario based on feedback from a clinical expert. The clinical expert suggested that the utility between the first and second progression for mirvetuximab is expected to be equal to the pre-progression utility for chemotherapy in MIRASOL. This is because people who have progressed after mirvetuximab would have the same chemotherapy treatments for their next treatment line as people having comparator chemotherapy. In its review of the company's draft guidance response, the EAG recalled the results of the company's original utility regression analysis. It noted that this did not include a treatment interaction effect and produced a small difference of around 0.03 between treatment arms, both pre- and post-progression. The EAG added that the company's updated utility regression analysis, including the treatment interaction effect, resulted in a smaller difference of around 0.02 pre-progression and a much larger difference of around 0.08 post-progression. The EAG considered this to lack face validity. It noted that, based on clinical and patient expert feedback, most of mirvetuximab's benefit would be expected before progression while people were having treatment. This is because mirvetuximab is expected to reduce the side effect burden and delay the time to the next treatment line. The EAG noted that the large post-progression utility benefit (nearly 4 times greater than pre-progression utility benefit) was difficult to justify. This was even after accounting for the increased time to second progression and increased chance of responding to subsequent chemotherapy. The EAG also noted that the unexpected results of the company's updated analysis, with the interaction term, may have been influenced by low sample size and high levels of missingness post-progression. It noted that the mixed model for repeated measures method used by the company assumed data was missing at random, but no justification for this assumption was given. It added that the sensitivity analyses in the company's draft guidance addendum to explore the impact of this missingness did not account for missingness across the length of follow-up for post-progression observations. It remained unclear to the EAG whether the mixed model for repeated measures was appropriate under the missing-at-random assumption. Regarding the company's scenario, the EAG agreed with the logic that quality of life will be, at least partly, dependent on the type of treatment people are having. But it noted that the same treatments are available after mirvetuximab and chemotherapy and so a large differentiation in post-progression utilities between treatment arms was unlikely to be justifiable. The EAG also noted that the company could have reanalysed its trial data to identify whether post-progression utility values were observed before or after the second progression, rather than basing this scenario analysis on assumptions. The committee agreed with the EAG that the results of the company's updated utility regression analysis, including the interaction term, lacked face validity. It also agreed that the company's scenario analysis could have been based on trial data. So, it did not believe the company's additional analyses to be methodologically robust or appropriate to use in the model. The committee concluded that the additional evidence presented was not sufficient to change its decision since the first committee meeting. The committee maintained that it preferred to use MIRASOL utilities without the interaction term.

In response to the draft guidance consultation, patient group surveys highlighted the substantial burden of platinum-resistant ovarian cancer, and its treatment, on unpaid carers. They explained that people with the condition rely heavily on others for support with basic tasks because of the high symptom burden and debilitating side effects of chemotherapy. During the second committee meeting, a clinical expert explained that people with ovarian cancer have a high symptom burden and high care needs in the last 12 months of life, particularly if they are older and have comorbidities. A patient expert explained that people's care needs are often related to the effects of peripheral neuropathy, a side effect of chemotherapy. They added that, for the people who had had mirvetuximab, treatment was better tolerated and there was less of a burden on carers. A clinical expert agreed that they would expect the carer burden to be reduced with mirvetuximab. This is because people having mirvetuximab may not need as many hospital visits compared with people having paclitaxel chemotherapy (which may be administered weekly). They added that people having mirvetuximab may also not need as many hospital visits for parenteral nutrition for bowel obstruction compared with people having paclitaxel or PLD.

In response to the draft guidance consultation, the company provided evidence from a study that explored the carer support needed by people with platinum-resistant ovarian cancer across Canada, France, Italy, Spain and the UK (Adelphi Real-World Evidence Survey). Based on this study, the company noted that carers of people with the condition spent an average of 35.4 hours per week providing care. The company said that this was an uncaptured benefit in the model. In its review of the company's draft guidance response, the EAG noted that the Adelphi Real-World Evidence Survey indicated that 50% of respondents needed additional support or care and 11% had a formal, rather than an informal carer. Based on this, the EAG estimated that 39% of people had an informal carer. The EAG said that it had used this estimate, alongside a study by Pennington et al. (2025), to estimate carer quality-adjusted life year (QALY) loss. During the second committee meeting, the EAG emphasised that this was an exploratory scenario analysis only, and was done to test the model's sensitivity to this assumption. The EAG noted that it did not consider the Adelphi Real-World Evidence Survey to be robust. It also noted that Pennington et al. was not specific to ovarian cancer. The committee noted the limitations of the EAG's exploratory scenario analysis. It added that it was unclear on how the EAG had used the Pennington et al. study to estimate the QALY loss for carers because the study considers utility and HRQoL only. The committee noted the expert's comments around reduced travel because of fewer hospital visits for people having mirvetuximab. It noted that section 4.3.17 of NICE's manual on technology appraisal and highly specialised technologies guidance explains that when presenting health effects for carers, evidence should be provided to show that the condition is associated with a substantial effect on carer's HRQoL. It did not consider that reduced travel would substantially contribute to an impact on carers' health. The committee acknowledged that platinum-resistant ovarian cancer had a similarly substantial burden on people with the condition and their carers as other types of advanced cancer. But it concluded that it had not seen robust evidence of the size of the carer burden or the role of mirvetuximab in alleviating carer burden. So, it concluded that a carer disutility should not be included.

At the first committee meeting, the company and EAG had differing approaches to modelling adverse events for the following assumptions:

• The company assumed the duration of grade 2 or higher ocular adverse events to be 4 weeks in its base case. But the EAG thought ocular adverse events were unlikely to resolve in 4 weeks and preferred to assume 8 weeks.

• The company assumed the frequency of ophthalmology visits for grade 2 or higher ocular adverse events to be once every 6 weeks, but the EAG preferred to assume once every 3 weeks.

• For the costs of managing anaemia and neutropenia, the company preferred to use a weighted average of day case, non-elective inpatient long-stay and non-elective inpatient short stay. This was because some people develop sepsis that needs to be managed in hospital. The EAG preferred to assume that anaemia and neutropenia were managed as day cases only.

• The company preferred to assume that parenteral nutrition was needed to manage fatigue, but the EAG preferred to exclude this cost and assume that fatigue is self-managed.
  
  The clinical experts said that grade 2 or higher ocular adverse events resolve relatively quickly. So, it would be more reasonable to assume people have an ophthalmology visit every 6 weeks instead of every 3 weeks. They also said that, while some people with neutropenia develop sepsis that needs managing in hospital, this happens rarely. The clinical experts also said that fatigue is managed with parenteral nutrition only in highly severe cases. Based on feedback from the clinical experts at the first committee meeting, the committee preferred to assume:

• a duration of 4 weeks for grade 2 or higher ocular adverse events

• a frequency of every 6 weeks for ophthalmology visits

• that anaemia and neutropenia is managed as a day case

• that fatigue is self-managed.
  
  In response to the draft guidance consultation, the company updated its base case to align with the committee's preference that fatigue is self-managed. But it maintained that a weighted average of day case, non-elective inpatient long-stay and non-elective inpatient short-stay costs for managing anaemia and neutropenia was more appropriate than assuming these were managed as day cases only. The company noted that this was consistent with the approach used in previous NICE evaluations of ovarian cancer treatments. In its review of the company's draft guidance response, the EAG explained that the company's weighted average cost was taken from NHS reference costs. So it reflected all diseases, rather than only being representative of chemotherapy side effects in platinum-resistant ovarian cancer. During the second committee meeting, 1 clinical expert explained that Hospital Episode Statistics data is available specifically for people having treatment for platinum-resistant ovarian cancer. The clinical expert noted that around 12% of neutropenia events were managed as inpatient stays. The committee concluded that 12% of neutropenia cases in the model should apply the cost of an inpatient stay, with the remainder of neutropenia events and anaemia events treated as day cases.

The company used a single average relative dose intensity (RDI) value across all cycles in the model for mirvetuximab and chemotherapy treatments. The EAG preferred to use a cycle-specific RDI that accounted for average RDI in the proportion of people having treatment in each model cycle. The EAG said that this approach allowed for a more accurate estimate of drug use and wastage. In response to the draft guidance consultation, the company maintained that its approach of using the average RDI was more appropriate than the EAG's cycle-specific approach. It presented a graph showing RDI over time and noted that the low patient numbers at later timepoints in the EAG's approach meant that this approach was not necessarily more accurate. In its review of the company's draft guidance response, the EAG explained that the company's average RDI for mirvetuximab was actually higher (less favourable to mirvetuximab) than long-term RDI in the EAG's cycle-specific analysis. The EAG explained that the reason its approach increases the ICER, is that in earlier cycles where patient numbers are highest, the trial-observed RDI values are greater than the mean. The committee noted that the graph presented by the company showed the RDI changing over time. It also noted that where there were most people on treatment, the RDI was above average. The committee concluded that the EAG's cycle-specific RDI approach more accurately reflected drug use and wastage in the model.

The company assumed 50% vial sharing for mirvetuximab and pooled chemotherapy. It noted that, in NICE's technology appraisal guidance on trastuzumab deruxtecan for treating HER2-positive unresectable or metastatic breast cancer after 1 or more anti-HER2 treatments, the committee concluded that vial sharing should be assumed in 50% of cases for trastuzumab deruxtecan. This was because the Cancer Drugs Fund lead had said that NHS England encourages vial sharing. The company thought that a similar degree of vial sharing may be expected for mirvetuximab. It also noted that its clinical experts said that vial sharing was common for PLD and paclitaxel. The EAG preferred to assume no vial sharing for mirvetuximab and 50% vial sharing for pooled chemotherapy. This was because its clinical experts were uncertain about the plausibility of vial sharing for mirvetuximab. They said that it would not be reasonable to delay treatment with mirvetuximab so that vials could be shared. The Cancer Drugs Fund lead agreed that vial sharing for mirvetuximab was unlikely to be feasible. They noted that the anticipated eligible population for mirvetuximab was substantially lower than for trastuzumab deruxtecan. So it was not reasonable to inform feasibility of vial sharing with mirvetuximab from vial sharing that occurs with trastuzumab deruxtecan.

In response to the draft guidance consultation, the company maintained that 50% vial sharing was appropriate for mirvetuximab. The company said that feedback from NHS pharmacists and consultants was that people with platinum-resistant ovarian cancer are frequently scheduled to have treatment on specific days of the week, so vial sharing would be possible. The company reiterated that they expect similar numbers of people to have mirvetuximab as are having trastuzumab deruxtecan. So, a 50% vial sharing assumption is appropriate. During the second committee meeting, the Cancer Drugs Fund lead noted that around 600 people start trastuzumab deruxtecan each year and people stay on treatment for 18 months on average. They explained that the patient numbers are lower and the time on treatment substantially shorter for mirvetuximab. This leads to substantially fewer people having mirvetuximab than trastuzumab deruxtecan at any time. This would be expected to result in fewer opportunities for vial sharing compared with trastuzumab. A clinical expert at the second committee meeting explained that ovarian cancer is treated in specialist centres with gynaecological clinics scheduled on the same day. They added that even if 2 people are scheduled for treatment on the same day, there is still the potential for cost savings through vial sharing. They were uncertain of the appropriate percentage to assume for vial sharing for mirvetuximab. They noted it would not be 50% but would also not be 0%. Another clinical expert stated that an assumption of 10% to 15% vial sharing was reasonable. This was aligned with an online comment received in response to consultation that said that an assumption of 10% to 15% vial sharing was achievable. The Cancer Drugs Fund lead agreed that some centres can group people having the same treatment, but this is the exception rather than the rule. They said that vial sharing for mirvetuximab was probably not 0% but not expected to exceed 10% to 15%. Based on these estimates, the committee concluded that the model should assume 10% vial sharing for mirvetuximab and 50% vial sharing for pooled chemotherapy.

The company sourced its resource-use frequencies from NICE's technology appraisal guidance on bevacizumab in combination with gemcitabine and carboplatin for treating the first recurrence of platinum-sensitive advanced ovarian cancer. It assumed that people in the pre-progression state had a gynaecological oncology consultation once monthly. In the post-progression state, it assumed that people had a gynaecological oncology consultation once every 3 months. The EAG's clinical experts said that visit frequency depended on whether a person was having treatment or not. They explained that a gynaecological oncology consultation took place once monthly while a person was on treatment and once every 3 months while they were off treatment. The EAG thought that the company's resource-use frequencies in the pre-progression state were broadly consistent with its clinical expectation. But, in the post-progression state, it preferred to assume that people had a gynaecological oncology consultation once monthly for the average duration of post-progression chemotherapy and then once every 3 months. During the committee meeting, the clinical experts explained that they tend to see people with the condition around once every 6 weeks. This is regardless of whether they are on or off treatment, or whether their cancer has progressed. Based on the advice from the clinical experts, the committee concluded that the model should include gynaecological oncology consultations once every 6 weeks in both the pre-progression and post-progression health states. The company updated its base case to align with the committee's preferred assumption at the draft guidance consultation.

The company modelled duration of treatment for mirvetuximab by fitting an exponential distribution to the duration-of-treatment curve from MIRASOL. It used the exponential distribution because it had the best statistical fit. The EAG thought that the company's approach underestimated duration of treatment for mirvetuximab, especially in the early part of the Kaplan–Meier curve. It also noted that the data for duration of treatment was complete, so it preferred to use the observed Kaplan–Meier data directly. The company said that using the Kaplan–Meier data risked overinterpreting small variations in the trial data. The committee noted that this issue had a small impact on the ICER. It agreed that both the company's and EAG's approaches were acceptable. But it noted that using a fitted distribution was more generalisable to clinical practice because it was less likely to overinterpret small changes in the Kaplan–Meier data. So, it concluded that it preferred the company's approach, which used the exponential distribution for modelling duration of treatment for mirvetuximab.

In its base case, the company included mirvetuximab as a subsequent treatment for 9% of the people in the chemotherapy arm who had a subsequent treatment. This was based on the treatment crossover in MIRASOL. It also provided a scenario analysis that used a rank-preserving structural-failure time model to adjust for the treatment crossover and removed the mirvetuximab costs post-progression in the pooled chemotherapy arm. The EAG thought that the inclusion of mirvetuximab as a subsequent treatment in the chemotherapy arm was inappropriate because it is not available in the NHS. It preferred to use the company's scenario analysis in its base case and removed the cost of mirvetuximab post-progression from both arms. The committee concluded that the EAG's approach was more appropriate. The company updated its base case to align with the committee's preferred assumption at draft guidance consultation.

Given the available evidence, the committee concluded that its preferred assumptions for the cost-effectiveness modelling were:

• using pooled chemotherapy, including PLD and paclitaxel, as the relevant comparator (see section 3.3)

• using the log-logistic distribution fit to SACT data for modelling pooled chemotherapy OS, with the MIRASOL OS HR of 0.68 applied to estimate OS for mirvetuximab (see section 3.10)

• applying the treatment-dependent health-state utility values from MIRASOL, without an interaction term, for both the mirvetuximab and pooled chemotherapy arms (see section 3.11)

• not including a carer disutility (see section 3.12)

• including disutilities associated with adverse events (see section 3.11)

• assuming the duration of grade 2 or higher ocular adverse events is 4 weeks (see section 3.13)

• assuming that the frequency of ophthalmology visits is every 6 weeks (see section 3.13)

• assuming that anaemia events are managed as day cases and 12% of neutropenia events require an inpatient stay (see section 3.13)

• assuming that fatigue is self-managed (see section 3.13)

• using the cycle-specific approach for modelling RDI (see section 3.14)

• assuming 10% vial sharing for mirvetuximab and 50% vial sharing for pooled chemotherapy (see section 3.15)

• having gynaecological oncology consultations once every 6 weeks in both the pre-progression and post-progression health states (see section 3.16)

• using the exponential distribution for modelling mirvetuximab duration of treatment (see section 3.17)

• adjusting for the treatment crossover and removing the cost of mirvetuximab in the post-progression state (see section 3.18)

• using the mean baseline age in SACT (67 years) for the starting age in the model (see section 3.20)

• applying a severity weight of 1.2 to the QALYs gained (see section 3.20)

• including the costs for baseline ocular assessment (see section 3.19)

• using the SB13Z HRG code for the first administration of paclitaxel in each 28-day cycle (see section 3.19)

• exploring both the EAG's and company's disutilities for alopecia (see section 3.19).

NICE's manual on technology appraisal and highly specialised technologies guidance notes that, above a most plausible ICER of £25,000 per 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 and health inequalities.

The committee thought that the SACT dataset was a robust and generalisable source of evidence for pooled chemotherapy. But, it also noted the uncertainty with using its preferred SACT plus HR approach to estimate long-term survival for mirvetuximab (see section 3.10). It noted that there were multiple plausible approaches that could be used for modelling OS (see section 3.10), and the choice of approach had a significant impact on the ICER. But the committee also acknowledged that there was a substantial unmet need for new treatments for platinum-resistant ovarian cancer and so mirvetuximab provided a novel treatment option for this condition. The committee noted the potential health inequality issues (see section 3.23) but it had not seen any evidence that mirvetuximab would substantially impact these. So, the committee concluded that an acceptable ICER would be around the middle of the range NICE considers a cost-effective use of NHS resources (£25,000 to £35,000 per QALY gained).

Stakeholders noted several equality issues:

• Ovarian cancer affects women, trans men and non-binary people registered female at birth.

• People from Caribbean and African backgrounds, people who are older and people from low-income groups may be more likely to be diagnosed with ovarian cancer at a later stage.

• Some groups of people (for example, people with diabetes) are at higher risk of developing severe peripheral neuropathy, making them ineligible for paclitaxel treatment.

• Mirvetuximab may require fewer hospital visits than paclitaxel. The simplified administration process may decrease travel time, which may result in financial savings for people having treatment and increase productivity.

• Adverse events of treatment may be more likely to impact people with language difficulties, or people from ethnic minority backgrounds who are more likely to have had poor experiences with healthcare.
  
  The committee noted that ovarian cancer affects people registered female at birth. It noted that people from Caribbean and African backgrounds and of an older age may be more likely to be diagnosed at a later stage. It recognised that sex and age are protected characteristics and ethnic background relates to the protected characteristic of race. But, it considered that this was an issue of prevalence within the licensed population. It noted that ethnic background, in combination with sex, age and income status, are issues of health distribution and unfair and sometimes avoidable differences based on diagnosis. The committee recognised these issues but considered that it had not seen any evidence that mirvetuximab would substantially affect the stage of diagnosis. It noted the time savings that mirvetuximab could offer, but noted that productivity benefits are not included in the NICE reference case. It added that the QALY benefits of less frequent administration should already be captured in the QALY calculation. The committee noted that some people may be ineligible for paclitaxel treatment. It noted that some characteristics that make people ineligible for paclitaxel treatment may be disabilities, which are protected under the Equality Act 2010. But, it noted that it had not seen any analyses that were specific to people who were ineligible for paclitaxel treatment. The committee concluded that because its recommendation does not restrict access to treatment for some people over others, there were no equalities issues that could be addressed within this evaluation.

The committee concluded that, compared with chemotherapy, mirvetuximab increases PFS and OS and improves the HRQoL of people with platinum-resistant ovarian cancer. When considering the condition's severity, and its effect on quality and length of life, the most likely cost-effectiveness estimates are within the range that NICE considers an acceptable use of NHS resources. So, mirvetuximab soravtansine can be used.