Topotecan, pegylated liposomal doxorubicin hydrochloride, paclitaxel, trabectedin and gemcitabine for treating recurrent ovarian cancer

4 Committee Discussion

The appraisal committee reviewed the data available on the clinical and cost effectiveness of topotecan, pegylated liposomal doxorubicin hydrochloride (PLDH), paclitaxel, trabectedin and gemcitabine, having considered evidence on the nature of recurrent ovarian cancer and the value placed on the benefits of these technologies by women with the condition, those who represent them, and clinical experts. It also took into account the effective use of NHS resources.

4.1 The committee heard from clinical experts that women with recurrent ovarian cancer can experience several relapses after initial treatment with platinum‑based chemotherapy, and it was very important to have a range of treatment options available at each relapse. The clinical experts stated that treatment is tailored to individuals, taking into account factors such as previous treatment, and the potential for developing platinum resistance. The patient experts stated that there is no screening programme for ovarian cancer and disease is usually identified at an advanced stage. They highlighted the emotional impact of developing recurrent ovarian cancer, particularly emphasising the fear of further recurrence and the great importance of progression‑free survival to patients' wellbeing. They also highlighted the psychological benefit of having a range of treatment options available. The clinical experts stated that ovarian cancer is increasingly seen as a group of diseases, that histological subtype plays an important role in how the disease responds to particular treatments, and that many treatments are not very effective in the rarer histological subtypes. Therefore, a range of chemotherapy agents is needed until more targeted therapies become available. The committee heard that approximately 70% of people with ovarian cancer have serous adenocarcinoma, and this was not expected to vary significantly across the trials included in the review. The committee concluded that progression‑free survival was an important outcome measure and noted that availability of a range of treatment options is valuable for treating recurrent ovarian cancer.

4.2 The committee discussed current clinical practice for treating recurrent ovarian cancer. The committee heard from the clinical experts that the assessment group's approach of presenting results separately for women with platinum‑sensitive disease and platinum‑resistant or platinum‑refractory disease was appropriate. The committee heard that the majority of patients in clinical practice have platinum‑sensitive disease at first recurrence, but this proportion would decline at each subsequent recurrence. The clinical experts stated that standard treatment for women with platinum‑sensitive disease, including those with partially platinum‑sensitive disease, is platinum‑combination chemotherapy. The clinical experts pointed out that the platinum‑resistant and platinum‑refractory group is heterogeneous because it includes women whose disease may never have responded to platinum as well as women whose disease developed resistance over time. The committee noted that no trials had taken this into account. The committee heard from the clinical experts that only a small proportion of women are allergic to a particular platinum agent such as carboplatin, and that these women are offered either an alternative platinum agent (cisplatin) or non‑platinum regimens. Desensitisation could also be carried out. The committee considered the assessment group's assumption that women with a platinum allergy would have the same probability of response to non‑platinum regimens as women without an allergy to be appropriate. The committee concluded that the assessment group's approach to the decision problem was appropriate.

4.3 The clinical experts stated that in clinical practice, the licensed dose may not be adhered to. For example, the licensed 3‑weekly paclitaxel regimen is used for platinum‑sensitive disease, but it is established clinical practice to use weekly paclitaxel for treating platinum‑refractory or platinum‑resistant disease. The committee was aware that the trials in platinum sensitive disease used 3‑weekly paclitaxel. One of the trials identified by the assessment group in platinum refractory/resistant disease (Lorthoraly et al.) used a weekly paclitaxel regimen. Although this trial had not been included in network 3, the experts consulted by the assessment group and the clinical experts at the appraisal committee meeting all confirmed that this was now standard practice in platinum‑refractory or resistant disease. The committee also heard that in practice, PLDH monotherapy is usually used at a lower dose than specified in the licence, and that the 40 mg/m² dose had become standard practice. For platinum‑sensitive disease in people who could not take platinum, the committee noted that one of the trials in network 2 used a dose of 45 mg/m² (Bafaloukos et al.). The committee concluded that in clinical practice the licensed dose is frequently adjusted to balance efficacy against toxicity, and that there was strong clinical opinion that using 40 mg/m² PLDH as monotherapy in platinum sensitive or resistant disease, or a weekly regimen of paclitaxel in platinum‑resistant or refractory disease does not reduce the clinical effectiveness of the treatments.

4.4 The committee discussed the clinical‑effectiveness evidence available, focusing on results from the assessment group's network meta‑analyses. The committee noted the following:

4.5 The committee discussed the limitations of the analysis, particularly the differences in baseline characteristics between trials, uncertainty about whether trials were adequately powered to detect differences in overall survival and progression‑free survival, and concerns about confounding because of crossover. However, the committee also heard from the clinical experts that the results of the network meta‑analyses were broadly in line with those expected from both the trial data and experience in clinical practice. The committee acknowledged that the analyses had methodological limitations and that some assumptions had to be accepted. However, it noted that they had been constructed using hazard ratios from peer reviewed publications, and represented a distillation and systematic analysis of the same body of evidence that is used by clinicians and patients when deciding between the various treatments. The committee expressed disappointment with the quality and breadth of the trial evidence, and also with some of the trial design and reporting, but on balance agreed that the assessment group's approach was reasonable given the data available, and accepted the clinical‑effectiveness results from the network meta‑analyses. It concluded that for women with platinum‑sensitive disease, paclitaxel, gemcitabine and PLDH (all plus carboplatin) improved progression‑free survival compared with platinum alone and that PLDH and paclitaxel (both plus carboplatin) also improved overall survival. It also concluded that trabectedin plus PLDH improved progression‑free survival compared with PLDH, paclitaxel and topotecan (all given alone), and that there was evidence that PLDH alone and trabectedin plus PLDH increased overall survival compared with topotecan. Finally, it accepted that for women with platinum‑resistant or platinum‑refractory disease, there were no statistically significant differences in progression‑free and overall survival between PLDH, paclitaxel and topotecan.

4.6 The committee discussed the cost‑effectiveness analyses conducted by the assessment group. It considered the cost‑effectiveness results based on network 1 for women with platinum‑sensitive recurrent ovarian cancer receiving platinum‑based chemotherapy, and noted that the fully incremental deterministic results indicated that paclitaxel plus platinum was the most cost‑effective treatment, with an incremental cost‑effectiveness ratio (ICER) of £24,400 per quality‑adjusted life year (QALY) gained compared with platinum alone. The committee also noted that although PLDH plus platinum was dominated by paclitaxel plus platinum and was therefore excluded from the fully incremental analysis, the costs and QALYs were very similar to those of paclitaxel plus platinum, and the ICER for PLDH plus platinum compared with platinum alone was approximately £30,200 per QALY gained. The committee concluded that paclitaxel plus platinum was the most cost‑effective option for women with recurrent platinum‑sensitive ovarian cancer but that PLDH plus platinum could also be considered a cost‑effective alternative.

4.7 After an appeal against its provisional guidance, the committee revisited its recommendation for using PLDH in combination with carboplatin. It was aware that use of this combination, based on the CALYPSO trial, requires giving PLDH at a dose lower than that specified in the marketing authorisation for PLDH (Caelyx), but at a dose equivalent to its licence for use in combination with trabectedin. During the appraisal, NICE had requested an exceptional direction from the Department of Health to make a recommendation for a drug outside the terms of its marketing authorisation. The committee had expressed concern that without such a direction, it would only be able to recommend a combination of paclitaxel and platinum for people with platinum‑sensitive disease, and that paclitaxel would not be suitable for some people because of toxicity or intolerance. In addition, the combination of PLDH with platinum is both clinically and cost effective, and is currently in clinical use. After NICE's request, the Department of Health obtained Ministerial agreement to direct NICE to make a recommendation under regulation 5 of the National Institute for Health and Care Excellence (Constitution and Functions) and the Health and Social Care Information Centre (Functions) Regulations 2013. The committee understood that this exceptional direction was given in the light of the committee's appraisal of the evidence, and took into account the need for a range of options for treating platinum‑sensitive recurrent disease. The committee was aware that under regulation 5, its recommendation would not be associated with mandatory funding and this was stated in the final appraisal determination issued to consultees for appeal. However, prolonged negotiations had delayed release of the document, and the wording in the final appraisal determination differed from that in the provisional recommendations in the appraisal consultation document. The appeal panel found that it was unfair not to have communicated transparently and provided an opportunity for consultees and commentators to respond to the proposed recommendation for the use of PLDH with platinum. On hearing the appeal panel's findings, the committee agreed that its recommendations were appropriate, and should stand, but that in line with the appeal panel's decision, consultees and commentators should have a further opportunity to comment on the document. Following consultation, in which no comments were received on the provisional recommendation for PLDH in combination with carboplatin, the committee concluded that the recommendation was appropriate.

4.8 The committee discussed the cost‑effectiveness results for gemcitabine plus carboplatin based on network 1. It noted that gemcitabine plus carboplatin was extendedly dominated and was therefore excluded from the incremental analysis including all the treatment options. It also noted that that the cost‑effectiveness estimates based on the network 1 meta‑analysis showed that gemcitabine plus carboplatin resulted in an ICER compared with platinum alone of £114,000 per QALY gained. The committee noted that the results of the economic model based on the network 1 meta‑analysis showed that gemcitabine plus carboplatin produced fewer QALY gains, representing less clinical benefit than paclitaxel plus platinum and PLDH plus platinum. The committee considered that this was primarily because of the lack of overall survival benefit demonstrated with gemcitabine plus carboplatin in the study by Pfisterer et al. The committee noted that this trial was powered to detect a statistically significant difference in progression‑free survival between treatments but not to detect a statistically significant difference in overall survival. It also noted statements from one of the clinical experts that the trial took place at a time when few post‑progression therapies were available, and that overall survival has since improved because patients are now offered multiple lines of therapy. The committee noted that extended follow‑up data from Pfisterer et al. were not available. However if multiple lines of follow‑on therapies had been given, although this could have increased the overall survival in both arms, confounding from subsequent lines of therapy might make the comparison between gemcitabine plus platinum and platinum alone even less reliable. Following consultation, the committee also considered a suggestion that an overall survival advantage might not have been found because patients in the trial had received a prior taxane and might therefore have had more resistant disease than patients in other studies. The committee acknowledged that there were uncertainties but concluded that Pfisterer et al. was the only relevant randomised controlled trial of gemcitabine plus platinum compared with platinum alone on which it could base its recommendations. It was of reasonable size, properly conducted, and was therefore appropriate for decision‑making as part of the network 1 meta‑analysis. The trial had not shown a statistically significant improvement in overall survival for gemcitabine plus platinum compared with platinum alone, unlike the combinations of paclitaxel and PLDH with platinum. The committee agreed that any gain in progression‑free survival was important to patients, but was satisfied that the benefits of treatment with gemcitabine on progression‑free survival had been adequately captured in the model.

4.9 Although the committee was satisfied that the network meta‑analysis was suitable for decision‑making, it noted that the clinical evidence for gemcitabine plus platinum that informed the cost‑effectiveness analysis only included women with a first recurrence of ovarian cancer, at which stage paclitaxel plus platinum or PLDH plus platinum would be alternative treatment options. The committee acknowledged that the sensitivity analysis around overall survival estimates indicated a high degree of uncertainty, and it concluded that gemcitabine could not be considered a cost‑effective use of NHS resources for treating a first recurrence of platinum‑sensitive ovarian cancer.

4.10 The committee heard from the clinical experts that patients with ovarian cancer often had multiple lines of treatment, and that it was important to have a wide range of treatments available for use in future recurrences. The committee noted that there was no evidence included in the assessment report on the clinical effectiveness of gemcitabine for recurrences after the first. It noted that the marketing authorisation for gemcitabine did not explicitly limit it to first recurrence. Although the committee accepted the results of the network meta‑analysis, it considered that it could not assume that identical results for relative clinical effectiveness would be found in women who had a second or subsequent recurrence. The committee concluded that it could not make any recommendation about the clinical or cost effectiveness of gemcitabine beyond the first recurrence of platinum‑sensitive ovarian cancer, and that clinicians should take this into account when considering gemcitabine plus carboplatin as a treatment option.

4.11 The committee discussed the cost‑effectiveness results based on network 2 in women with platinum‑sensitive disease receiving non‑platinum‑based treatments. The committee noted that the ICER for PLDH monotherapy compared with paclitaxel monotherapy was approximately £23,700 per QALY gained. It also noted the assessment group's comments that the costs and QALYs associated with paclitaxel were similar to those of PLDH. The committee noted that in the base case analysis the cost of PLDH monotherapy was based on the assumption that it was given at a dose lower than the dose specified in the summary of product characteristics. The committee recalled that the clinical experts had confirmed that dose‑reduction was usual in clinical practice (see section 3.46), and that the clinical effectiveness at this dose was considered to be comparable with the trial evidence. However, it was also aware that the appeal panel required the appraisal committee to clarify and justify the reasoning for its choice of appropriate dose regimens for different treatments and the costs used in the cost‑effectiveness analyses. The committee therefore gave further consideration to its choice of the most appropriate costing for all treatments in the modelling of cost effectiveness. The committee discussed the 3 possibilities for estimating drug costs: the dose used in clinical practice in the NHS, where this was well established; the licensed dose; and the average doses used in the trials that informed the networks. The committee noted paragraph 5.5.1 from the NICE guide to the methods of technology appraisal 2013 which states that 'For the reference case, costs should relate to resources that are under the control of the NHS. These resources should be valued using the prices relevant to the NHS'. The committee noted that the assessment group had used the doses in clinical practice as the most relevant. The committee was minded to agree that for a complex appraisal of several different treatments in 3 different clinical scenarios, the dose used in clinical practice, where this was well established, was the most relevant for estimating the cost of treatment in the NHS. However, it reconsidered the other options. With regard to using the licensed doses for all treatments, and assuming adherence to the full licensed starting dose for the whole course of treatment, the committee noted that there was clear evidence that for some treatments this did not happen in clinical practice, and indeed the marketing authorisation allowed dose reduction in the case of toxicity. The committee concluded that calculating costs on the assumption of adherence to the full starting dose was less relevant to the NHS than the doses used in clinical practice. Regarding the use of average trial doses, several trials had been used in the networks to assess clinical effectiveness. The committee anticipated that the average dose received by patients in the trials would be lower than the licensed dose, because of toxicity, but estimates of the average dose were not uniformly available. Although the committee appreciated that the doses used in clinical practice, even when these are well established, may not be identical to the average doses in the trials, it was reassured that there was no clinical expectation that the effectiveness achieved by the treatments as currently administered in the NHS would be significantly different to the effectiveness demonstrated in the trials. The committee concluded that the doses used in clinical practice were generally agreed and consistent, and most relevant to the NHS. It therefore agreed with the assessment group, and its own previous conclusion, that cost estimates based on doses used in clinical practice were the most appropriate to use in the cost effectiveness analyses. Specifically for PLDH, the committee considered that the estimate of treatment cost based on the licensed dose was higher than would be achieved in clinical practice and consequently the ICER of £31,200 per QALY gained for PLDH compared with paclitaxel (see section 3.39) was too high and the base case estimate of the ICER (£23,700 per QALY gained, see section 3.37) was reasonable. The committee concluded that both paclitaxel and PLDH could be recommended for use in the NHS for women with platinum‑sensitive disease for whom platinum‑based treatment was unsuitable.

4.12 In network 2, topotecan produced the fewest QALYs compared with PLDH monotherapy, trabectedin plus PLDH, and paclitaxel. However, it was associated with higher costs than both PLDH and paclitaxel monotherapy and was therefore dominated and excluded from the fully incremental analysis. The committee agreed that in the network meta‑analysis topotecan had not been demonstrated to be cost effective for the treatment of platinum‑sensitive ovarian cancer in women unable to take platinum and that this remained the case in the sensitivity analyses. It understood that having a range of treatment options was desirable, although topotecan was not widely used in clinical practice. However, the committee also acknowledged that the bulk of the evidence for topotecan in the context of platinum‑sensitive disease was for women with a first recurrence. The committee concluded that topotecan is not a cost‑effective use of NHS resources for a first recurrence, but was unable to make a recommendation for the use of topotecan in platinum‑sensitive disease beyond the first recurrence.

4.13 The committee carefully considered the cost‑effectiveness results for trabectedin plus PLDH compared with PLDH alone, and the comments received from the company for trabectedin during consultation. It noted the company's comment that the assessment group had overestimated the cost of a course of trabectedin and PLDH, and underestimated the cost of a course of PLDH alone, because it had based its calculations on a dose that was lower than the one specified in the marketing authorisation and lower than the average dose received by patients in the OVA‑301 trial. The committee noted that it had accepted using the costs for a 40 mg/m² of body surface area dose for PLDH monotherapy (as used in clinical practice) as reasonable, but that if the licensed dose of PLDH monotherapy had been used, this would have reduced the ICER for trabectedin plus PLDH compared with PLDH alone by approximately £8,000 per QALY gained (see section 3.39), calculated using the network meta‑analysis for network 2. The committee concluded that, even with this adjustment, the ICER remained too high for it to recommend the combination of trabectedin plus PLDH as a cost‑effective use of NHS resources.

4.14 The committee understood that the company's economic evaluation, which used clinical‑effectiveness data obtained from the OVA‑301 trial, had been retrospectively adjusted for the following potential covariates: imbalances between arms in the platinum‑free interval, Eastern Cooperative Oncology Group (ECOG) status, and the antigen CA125. It noted that the resulting ICER using the company's model, based on the original patient access scheme, was £27,600 per QALY gained, and that this was substantially lower than the ICERs calculated by the assessment group based on network 2. The new base‑case ICER was also substantially lower than the ICER previously estimated by the company (over £94,800 per QALY gained [without the patient access scheme], using OVA‑301 data) in NICE technology appraisal guidance 222. The committee heard from the company that the substantial reduction in the ICER since NICE technology appraisal guidance 222 was predominantly because of the post‑hoc adjustment for imbalances in the platinum‑free interval that were subsequently discovered, and that pre‑specified adjustments to ECOG status and CA125 had less of an effect. The committee heard from the company that the results were also partly influenced by the choice of modelling distributions, which had changed from exponential distributions in NICE technology appraisal guidance 222 to Weibull and log logistic distributions in the current appraisal because more mature overall survival data were available. The committee accepted the views of the assessment group that the company had used the best‑fitting distributions in its model. However, the committee was concerned that the log logistic extrapolation resulted in 2% of women in the treatment arm being alive after 15 years, which it considered was likely to be optimistic for women with recurrent advanced ovarian cancer.

4.15 The committee carefully considered the company's adjustment of the treatment effects, noting that this increased the difference in median overall survival between the treatment groups from a non‑statistically significant 2.9 months to a statistically significant 4.3 months. The committee considered the company's consultation comment that the NICE Decision Support Unit guidance for survival analyses recommends 'adjusting for the characteristics of the patients included in the clinical trial of interest – thus correcting for any patient population differences which may be present between different clinical trials'. However, it understood from the assessment group that the focus of the Decision Support Unit guidance is on head‑to‑head comparisons in which patient‑level data are available and where evidence synthesis between trials is not required. Guidance on adjustment of dissimilar trial level data to create a homogeneous network for which there is only a single trial informing each head‑to‑head comparison is not available. The committee concluded that it was not required to accept the suggested post hoc adjustments of the trial data.

4.16 The committee also considered the company's concerns that the conclusions around retrospective adjustment of treatment effects in the appraisal consultation document were inconsistent with those made in NICE technology appraisal guidance 222. The committee was aware that the Evidence Review Group for NICE technology appraisal guidance 222 had accepted an adjustment for CA125, and so the committee asked what the ICER would have been if the company had adjusted for CA125 alone in the current appraisal. The company indicated that this analysis was not available, but that CA125 adjustment did not have a major effect on the ICER. The committee also noted that the appraisal committee for NICE technology appraisal guidance 222 had not made an explicit judgement about the validity of the retrospective adjustment of CA125, but had favoured the Evidence Review Group's analysis because it used data from the fully platinum‑sensitive population rather than from a post hoc partially sensitive subgroup as in the company's submission. The committee concluded that it would have given consideration to a retrospective adjustment for CA125 for consistency with NICE technology appraisal guidance 222, had it been supplied by the company, but that this did not oblige the committee to accept additional retrospective adjustments.

4.17 The committee examined the assumption underpinning the adjustment for platinum‑free interval, that is that response to all future therapy would be better the longer the platinum‑free interval. The committee questioned whether this applied only to further platinum therapy. One of the clinical experts indicated that the platinum‑free interval was important, subject to ongoing research, and that a longer platinum‑free interval may be associated with a better response to non‑platinum therapies as well. The clinical expert stated that no trials have yet provided evidence to support this hypothesis for PLDH, however the committee noted that the company considered that the post‑hoc analyses from OVA‑301 supported this hypothesis.

4.18 The committee also considered the company's approach to carrying out the retrospective analysis. It was concerned that the platinum‑free interval had been treated as a continuous rather than a categorical variable. It expressed concerns that this had not been done for any other trials, and questioned the accuracy with which the platinum‑free interval could be assessed as a continuous variable, because assessments are made at set intervals, and a precise date of progression would not be known. The committee also questioned why the platinum‑free interval had not been a stratification factor at randomisation (except for with the platinum‑sensitive and platinum‑resistant groups) if it was critical to the interpretation of the trial results. The company responded that the platinum‑free interval was not considered as important a prognostic indicator at the time of the trial. It acknowledged that it had not included adjustments for the platinum‑free interval in its submission for the whole platinum‑sensitive population for NICE technology appraisal guidance 222 because the imbalance was not known until the final analysis of overall survival was available, which was after the final guidance from technology appraisal guidance 222. However, it was now of the opinion that it should now be retrospectively adjusted for. The committee agreed that the platinum‑free interval, recognised as an important prognostic factor, at least for further platinum therapy, should be considered carefully in the design and pre‑specified analyses of future ovarian cancer trials. However, the committee was concerned that retrospective adjustment for the platinum‑free interval in non‑platinum treatments, and as a continuous variable, is currently an unvalidated post‑hoc approach, and was not sufficiently reliable as a basis for estimating effectiveness.

4.19 The committee also evaluated the impact of the retrospective adjustments suggested by the company in which the platinum‑free interval was treated as a continuous variable. Using the company's adjusted results, and its new model, there was a calculated mean overall survival benefit of 9.7 months for trabectedin plus PLDH compared with PLDH alone. This contrasted with a mean 2.9 months overall survival gain as calculated by the assessment group. The committee did not consider a 9.7 month overall survival gain to be plausible, given that no statistically significant overall survival benefit was demonstrated in OVA‑301. The committee considered that the company's modelled overall survival benefit lacked credibility, and that this cast further doubt on the validity of the retrospective adjustment of treatment effects that had been introduced since the company's previous analyses presented for NICE technology appraisal guidance 222.

4.20 The committee carefully considered the additional analyses for trabectedin plus PLDH compared with PLDH alone, and the comments received from the company for trabectedin during consultation on the post‑appeal appraisal consultation document. It noted that the company had reiterated its view that the true benefit of trabectedin could only be captured if the results from the platinum‑sensitive subgroup of the OVA‑301 trial were retrospectively adjusted for imbalances in the platinum‑free interval, and they had submitted further analyses retrospectively adjusting for the platinum‑free interval as a categorical rather than a continuous variable. The committee re‑examined the evidence underpinning the company's adjustment. It noted that this was based on post‑hoc analyses derived from OVA‑301, using results presented in a 2012 publication by Monk et al. that was highlighted in the company's response. It agreed with the comments in the paper by Monk et al. that the analyses were 'hypothesis generating only', and that 'no firm conclusions could be made' because they were not pre‑specified, and that 'these ad‑hocexploratory analyses' required 'prospective validation'. It also agreed with the assessment group that results from a second, independent study would be needed to demonstrate an association between duration of the platinum‑free interval and response to non‑platinum based treatments. After further consideration, the committee confirmed its earlier view (see section 4.18) that the retrospective adjustment for platinum‑free interval as presented by the company is currently not sufficiently reliable as a basis for estimating effectiveness. Furthermore, the committee remained concerned that the adjusted analyses resulted in a substantially higher mean overall survival gain for trabectedin plus PLDH of 8.36 months, compared with the mean of 2.9 months calculated by the assessment group. The committee was not persuaded that the substantial overall survival benefit of 8.36 months was a reliable finding, given that no statistically significant overall survival benefit was demonstrated in OVA‑301. It concluded that the retrospectively adjusted results did not provide a more robust estimate of the clinical effectiveness of trabectedin plus PLDH than the published unadjusted results from the properly conducted and randomised OVA‑301 trial which had been incorporated into the assessment group's network meta‑analysis.

4.21 The committee considered the revised cost‑effectiveness results submitted by the company that incorporated the retrospective adjustments for imbalances in the platinum‑free interval as a categorical variable and an updated patient access scheme (see section 2.14). The committee was not persuaded that the ICERs based on the retrospective adjustment were robust for the reasons set out in section 4.20. It addition, it was concerned that adjusting for the platinum‑free interval led to substantially different ICERs depending on the method of adjustment (£28,600 per QALY gained in the company's analysis using patient level data and £59,800 per QALY gained in the assessment group's analysis using the company's adjusted hazard ratios, both incorporating the updated patient access scheme). The committee considered that this further undermined the plausibility of the analyses. It agreed that the assessment group's ICERs, based on the unadjusted results, gave a more plausible estimate of the cost effectiveness of trabectedin plus PLDH compared with PLDH alone. It noted that taking into account the updated patient access scheme, the assessment group's ICER remained above £70,000 per QALY gained. The committee was not convinced that the ICER for trabectedin plus PLDH fell within the range which could be considered a cost effective use of NHS resources.

4.22 The committee was aware that the company had also reiterated other concerns, including that cost estimates were based on doses used in clinical practice instead of average doses from the clinical trials. However, the committee maintained its view that this was appropriate because average doses were not uniformly available, and doses used in clinical practice are both generally agreed and consistent, and most relevant to the NHS (see section 4.8). Furthermore, it acknowledged that accounting for this had a minimal impact on the ICER. The committee concluded that the most plausible ICER remained above £50,000 per QALY gained even if it had accepted the additional issues raised by the company.

4.23 The committee noted that OVA‑301 included only women with a first recurrence of ovarian cancer, and no clinical‑effectiveness evidence for trabectedin was available for subsequent recurrences. It therefore concluded that trabectedin plus PLDH could not be considered a cost‑effective use of NHS resources for treating a first recurrence of platinum‑sensitive ovarian cancer. It further concluded that it could not make any recommendation about the clinical or cost effectiveness of trabectedin plus PLDH in later recurrences, and that clinicians should take this into account when considering trabectedin plus PLDH as a treatment option.

4.24 The committee then discussed network 3 and the cost‑effectiveness results for paclitaxel, PLDH and topotecan, all given as monotherapy for women with platinum‑refractory or platinum‑resistant disease. In the comparison between PLDH and paclitaxel, the committee noted that paclitaxel was dominated by PLDH and therefore excluded from the incremental analysis. However, the committee noted the clinical experts' opinion that paclitaxel was considered standard care in this setting, and that the costs and QALYs associated with paclitaxel were similar to those of PLDH (£900 difference in total costs and 0.022 difference in QALYs). The committee noted that the base case ICERs had been calculated on the assumption that PLDH would be given at a lower dose than specified in the marketing authorisation and that paclitaxel was assumed to be used in an unlicensed weekly regimen (80 mg/m²/week). It considered the additional analyses produced by the assessment group after appeal (see section 3.42). It noted that when the licensed 3‑weekly paclitaxel regimen was assumed, the costs associated with paclitaxel were lower and it was no longer dominated. The ICER for PLDH relative to paclitaxel was £69,900 per QALY gained at a PLDH dose of 40 mg/m² of body surface area and £103,800 per QALY gained at a PLDH dose of 50 mg/m² of body surface area. The committee considered that the substantial impact that varying the drug cost had on the ICER was because of small differences in costs and effects, resulting in unstable ICERs. It also considered that its previous opinion had been correct; that the best estimates of costs were those based on current clinical practice. The committee concluded that these analyses did not change its previous conclusion that the costs and QALYs associated with paclitaxel and PLDH were similar. The committee concluded that both paclitaxel and PLDH could be recommended for use in the NHS for women with platinum‑refractory or platinum‑resistant ovarian cancer.

4.25 The committee noted that the incremental ICER for topotecan compared with PLDH was approximately £450,000 per QALY gained and remained above £50,000 per QALY gained in the sensitivity analyses. The committee also noted the previous comments from the clinical experts that topotecan is rarely used in clinical practice for platinum‑resistant disease because of low response rates. The committee therefore concluded that topotecan could not be considered a cost‑effective use of NHS resources, for treating platinum‑resistant or ‑refractory ovarian cancer.

4.26 The committee considered supplementary advice from NICE that should be taken into account when appraising treatments that may extend the life of people with a short life expectancy and that are licensed for indications that affect small numbers of people with incurable illnesses. For this advice to be applied, all the following criteria must be met:

4.27 The committee was aware that the company for trabectedin had stated that using the adjusted trial results, the median life expectancy in the platinum‑sensitive population was 19.4 months, and a gain in median life expectancy of 4 months was estimated with trabectedin for the platinum‑sensitive and partially platinum‑sensitive populations. However, the committee did not accept that the adjusted trial results were sufficiently reliable as a basis for estimating effectiveness (see sections 4.14–4.19) and noted that estimates from OVA‑301 discussed in NICE technology appraisal guidance 222 continued to be relevant. The committee noted that the median overall survival for people treated with PLDH in the entire platinum‑sensitive population was more than 24 months, and that the overall survival gain was less than 3 months and not statistically significant. The committee concluded that trabectedin in combination with PLDH did not fulfil the criteria for being a life‑extending, end‑of‑life treatment and that, even if it had met the criteria, the cost‑effectiveness estimates (see section 4.8) remained outside the range usually considered a cost‑effective use of NHS resources.

4.28 The committee was aware of NICE's position statement on the Pharmaceutical Price Regulation Scheme (PPRS) 2014, and in particular the PPRS payment mechanism. It accepted the conclusion 'that the 2014 PPRS payment mechanism should not, as a matter of course, be regarded as a relevant consideration in its assessment of the cost effectiveness of branded medicines'. The committee heard nothing to suggest that there is any basis for taking a different view about the relevance of the PPRS to this appraisal. It therefore concluded that the PPRS payment mechanism was not relevant in considering the cost effectiveness of any of the technologies in this appraisal.