Ramucirumab for treating advanced gastric cancer or gastro–oesophageal junction adenocarcinoma previously treated with chemotherapy

The Committee discussed the nature and quality of evidence for the clinical effectiveness of ramucirumab in combination with paclitaxel (RAINBOW) and for ramucirumab monotherapy (REGARD). Regarding ramucirumab monotherapy, the Committee was aware that the trial population for REGARD was very similar to that of RAINBOW, and it heard from clinical experts that the only notable difference between the trials was the choice of comparator. It heard that REGARD started 1 year earlier (2009) than RAINBOW (2010) and that the difference in comparators reflected a change in clinical practice in that year. The Committee considered that there was some uncertainty in whether people in REGARD were eligible for paclitaxel. However, it was persuaded that overall the evidence was representative of the population included in the marketing authorisation for ramucirumab monotherapy. It also noted that there was no other evidence put forward for this population, and considered it likely that no further evidence for this population would become available in the future. Regarding the evidence for ramucirumab plus paclitaxel from the RAINBOW trial, the Committee was aware that the population in the study was in agreement with the marketing authorisation and that the Evidence Review Group (ERG) had considered the study to be a good‑quality, international, randomised controlled trial with low levels of uncertainty because of the mature overall‑survival and progression‑free‑survival data. The Committee concluded that the REGARD trial was suitable evidence on which it could base a decision on the clinical efficacy of ramucirumab monotherapy. It further concluded that the evidence from the RAINBOW trial was also appropriate for basing a decision on the clinical efficacy of ramucirumab plus paclitaxel.

The Committee considered the relative clinical‑effectiveness evidence for ramucirumab. It was aware of the overall‑survival and progression‑free‑survival outcomes for ramucirumab plus paclitaxel compared with placebo plus paclitaxel (see section 3.5) and for ramucirumab plus placebo compared with placebo plus best supportive care (see section 3.13). Regarding the clinical safety of ramucirumab, the Committee was aware that neutropenia was the most frequently reported serious adverse event, with a higher incidence in people who had ramucirumab. The Committee noted that the European public assessment report (EPAR) on the clinical safety of ramucirumab concluded that it was generally acceptable and in line with other similar treatments, and the Committee therefore considered that overall it did not have any particular safety concerns. The Committee concluded that ramucirumab plus paclitaxel provided an extension to life of 2.3 median months in overall survival with similar toxicity to placebo plus paclitaxel. It further concluded that ramucirumab plus best supportive care provided a median extension to life of 1.4 months compared with placebo plus best supportive care, and that it also offers an active treatment option for people for whom cytotoxic chemotherapy is not considered appropriate.

The Committee discussed the network meta‑analysis that had been carried out by the company. It noted that the mean additional survival gains using the results of the network meta‑analysis for ramucirumab plus paclitaxel were 4.13 and 6.03 months compared with docetaxel and best supportive care respectively. The Committee questioned why these survival gains were substantially higher than the median overall‑survival gains seen from the trial data. It considered comments from the ERG that the results of the network meta‑analysis should be interpreted with caution because of the heterogeneity of the included studies. The network was weakened by 2 crucial links. First, the Committee was aware that the network of evidence relied on the Hironaka et al. (2013) study that was in an entirely Japanese population where there is a national screening programme to diagnose the disease in the earlier stages. It also heard from the clinical experts that patients included in the Hironaka et al. study had much longer survival gains than are typically seen in UK clinical practice. The Committee noted comments from the company during consultation that there was no clear biological rationale for why the hazard ratio from the Hironaka et al. study (which relied on the relative rather than absolute treatment effect of paclitaxel compared with irinotecan) would be different in a western population. However, the Committee noted that no evidence had been presented to confirm that this would be the case, and so it was still of the opinion that there was significant uncertainty about using the results of this trial in the network meta‑analysis. Second, the network also relied on the Thuss‑Patience et al. (2011) study, which had few patients and was stopped early. The Committee noted comments from the company during consultation that the early termination of the trial was related to difficulties in recruitment, rather than efficacy, so the trial results should still be unbiased. However, the Committee, noting that the trial was based on a sample of only 40 randomised patients, considered the point estimates for the treatment effect to be associated with considerable uncertainty. The Committee discussed how network meta‑analysis can be useful to strengthen the evidence base, but noted that in this case there were multiple single trials between ramucirumab and docetaxel, each associated with increased uncertainty, and with no link to ramucirumab other than the RAINBOW trial. The Committee was also mindful of clinical expert statements (see section 4.3) that the choice between paclitaxel and docetaxel depends on physician and patient preference. The Committee was aware that NICE's guide to the methods of technology appraisal (section 5.2.12) states 'Data from head‑to‑head RCTs should be presented in the reference‑case analysis'. The Committee therefore considered there to be no requirement to accept the results of the network meta‑analysis, with its associated uncertainty, rather than directly relevant head‑to‑head data from a good‑quality, international, randomised controlled trial with mature overall‑survival and progression‑free‑survival data. The Committee concluded that for the basis of decision‑making, the results of the network meta‑analysis would not be used in preference to the RAINBOW trial data comparing ramucirumab plus paclitaxel with paclitaxel plus placebo.

The Committee considered the company's economic models for ramucirumab combination therapy and monotherapy. It noted that the structures were identical, but that they included different populations according to the indications for combination therapy and monotherapy in the marketing authorisation. The Committee was aware that the structure of the models was commonly used for cancer cost‑effectiveness analyses with pre‑progression, post‑progression and death states, and considered that this was appropriate. The Committee was further aware that the ERG had included the following adjustments to the model in its base case (see section 3.45):

• Corrected programming errors and errors relating to the price of docetaxel.

• Corrected the double counting of hospitalisations because Healthcare Resource Groups' (HRGs) codes referring to adverse events also take hospitalisations into account.

• Used region 1 data for stratifying length of hospitalisation stay and treatment stratification.

• Used body surface area and body weight based on region 1 data instead of the intention‑to‑treat population data.
  
  The Committee was aware that the company had confirmed the errors related to programming and the price of docetaxel at clarification stage, noting that the impact on the ICER was negligible. The Committee considered that the drug acquisition costs for ramucirumab and paclitaxel had been underestimated by the company because they were based on the average weight of all people in the RAINBOW trial, about one‑third of whom were Asian. It therefore agreed with the adjustments carried out by the ERG to use region 1 data for body surface area and body weight. It also agreed with the ERG's adjustments to correct for double counting of hospitalisations, and to adjust length of hospitalisation stay for region 1. The Committee noted that the ERG's adjustments resulted in ICERs compared with best supportive care of £129,400 and £188,100 per QALY gained for the combination‑therapy and monotherapy models respectively. The Committee concluded that the model submitted by the company was robust and suitable for the purposes of its decision‑making and that the ERG's suggested amendments to the model were appropriate.

The Committee considered the use of health‑state utility values in the model. Regarding the health‑state utility value for the pre‑progression health state, the Committee understood that the model used the baseline utility value adjusted for treatment response and adverse effects. It was aware that the company's approach assumed that this value remained constant throughout the time that a person is in the pre‑progression health state. The Committee discussed the alternative approach suggested by the ERG in a scenario analysis, in which it used the RAINBOW trial data from different time points during the pre‑progression period to avoid adjusting the trial data. The Committee expressed a preference for data that do not have to be manipulated and therefore agreed that the ERG's approach was reasonable. It noted that the impact of this scenario analysis was very small (decreasing the ERG's base‑case ICER by less than £50 per QALY gained). The Committee concluded that, although it made only a minor difference to the ICER, it had a preference for using the ERG's approach to modelling pre‑progression health‑state utility and that this was incorporated into its consideration of the most plausible ICER.

The Committee considered the use of parametric curves for estimating progression‑free survival. It understood that the company had used this approach because the 6‑weekly assessments caused a stepped curve, so parametric curves were used to incorporate interval censoring. It noted that this approach was inconsistent with the methods used for the combination‑therapy model to estimate overall survival, in which the company had used Kaplan–Meier data from the trial. The Committee was aware that the company had carried out sensitivity analyses for the combination‑therapy model using alternative parametric curves and that this had made little difference to the ICER. The Committee noted that the progression‑free‑survival data were mature and it considered that the company's approach to fitting parametric curves to estimate progression‑free survival rather than using Kaplan–Meier data from the trial was not necessary. It concluded that although it has a preference for using direct trial data when they are available, this did not influence the cost‑effectiveness analyses and therefore the company's approach to modelling progression‑free survival was considered acceptable.

The Committee understood that treatment with ramucirumab is continued until disease progression and therefore the assumptions around frequency of assessment of disease progression were potential drivers of cost effectiveness. This was because the total acquisition costs of the technology were dependent on this assumption. It understood that people in the trial, and therefore in the model, were assessed for progression every 6 weeks. The Committee heard from the clinical experts that this reflected UK clinical practice. The Committee concluded that the assumption of 6‑weekly assessments was therefore reasonable.

The Committee considered the most plausible ICER for the combination‑therapy model. The Committee was mindful of its previous conclusions that paclitaxel and docetaxel are both appropriate comparators for ramucirumab combination therapy (see section 4.3), that the overall survival and progression‑free survival results of the network meta‑analysis should not be used in preference to the direct head‑to‑head data from the RAINBOW trial (see section 4.6), and its preferred approach to the utility values for the pre‑progression health state (see section 4.8). On this basis, the Committee considered the most robust estimate was the ERG's exploratory analysis, which used RAINBOW trial data for ramucirumab plus paclitaxel compared with placebo plus paclitaxel, and which used utility values from RAINBOW data over different time points during the pre‑progression period. The Committee therefore concluded that the most plausible ICER for people with gastric cancer or gastro–oesophageal junction adenocarcinoma for whom treatment in combination with cytotoxic chemotherapy is appropriate was £408,200 per QALY gained (representing incremental costs of £35,100 and incremental QALYs of 0.09).

The Committee considered the most plausible ICER for the monotherapy model. The Committee was aware of its earlier conclusions during which it established that best supportive care was a valid comparator (see section 4.3), and in which it accepted the adjustments to the model made by the ERG (see section 4.7). The Committee therefore concluded that the most plausible ICER for people with gastric cancer or gastro–oesophageal junction adenocarcinoma for whom further cytotoxic chemotherapy is not appropriate was £188,100 per QALY gained (representing incremental costs of £22,500 and incremental QALYs of 0.12).

The Committee considered supplementary advice from NICE that should be taken into account when appraising treatments that may extend the life of patients 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:

• The treatment is indicated for people with a short life expectancy, normally less than 24 months.

• There is sufficient evidence to indicate that the treatment offers an extension to life, normally of at least an additional 3 months, compared with current NHS treatment.

• The treatment is licensed or otherwise indicated for small patient populations.
  
  In addition, when taking these criteria into account, the Committee must be persuaded that the estimates of the extension to life are robust and that the assumptions used in the reference case of the economic modelling are plausible, objective and robust.

The Committee noted the views of the company that ramucirumab combination therapy fulfils all 3 end‑of‑life criteria when compared with best supportive care or docetaxel. It considered whether ramucirumab met the first criterion for an end‑of‑life treatment that patients have a short life expectancy, normally less than 24 months. It noted that patients in the placebo plus best supportive care arm of the REGARD trial had a median overall survival of 3.8 months, and patients in the placebo plus paclitaxel arm of the RAINBOW trial had a median overall survival of 7.4 months. The Committee therefore concluded that the life expectancy for people with advanced gastric cancer or gastro–oesophageal junction adenocarcinoma was less than 24 months.

The Committee considered whether ramucirumab offered an extension to life of at least 3 months. It was aware that the company had considered this criterion had been met using results of the network meta‑analysis for overall survival gains compared with best supportive care (6.03 months), and compared with docetaxel (4.13 months). It was also aware that the company had stated during consultation that the overall‑survival gains of all relevant comparators should be considered when determining if the end‑of‑life criteria have been met. The Committee considered the survival gains for all relevant comparators (see section 4.3 for the discussion of relevant comparators), but it was aware of its previous conclusions about the network meta‑analysis, and considered that the overall‑survival gain from the network meta‑analysis for ramucirumab plus paclitaxel compared with docetaxel was not plausible, objective or robust. The Committee considered that the RAINBOW trial data for overall survival were mature, and therefore the modelled mean overall survival gain of 1.30 months compared with placebo plus paclitaxel using direct evidence from the RAINBOW trial was the most robust estimate available, and consequently the only estimate on which it could consider the criterion relating to extension to life of at least 3 months. The Committee therefore agreed that ramucirumab plus paclitaxel did not offer an extension to life of at least an additional 3 months. The Committee was aware that the company had only presented a case for ramucirumab combination therapy to fulfil the end‑of‑life criteria, and that the median survival gain of ramucirumab monotherapy was 1.4 months compared with placebo plus best supportive care from the REGARD trial. The Committee therefore considered ramucirumab monotherapy did not offer an extension to life of at least an additional 3 months.

The Committee considered the total size of the population in the ramucirumab marketing authorisation for advanced gastric cancer or gastro–oesophageal junction adenocarcinoma for people who have already had platinum therapy. It noted that the company had estimated the size of this population in England as 657 people. It understood that the company had calculated this figure by using 2012 Office for National Statistics (ONS) data adjusted to 2015 (the effect of which was to reduce the figure by around 500 people) to give 8270 people with gastric cancer or gastro–oesophageal junction adenocarcinoma. The Committee heard from the ERG that since this appraisal started, 2013 ONS data have become available, and that these data showed the number of people with gastric cancer or gastro–oesophageal junction adenocarcinoma had not gone down, but had increased to 9198 people. The Committee therefore considered that any downward adjustment was not needed. The Committee agreed with the company, based on data from Cancer Research UK, that 80% of people with gastric cancer or gastro–oesophageal junction adenocarcinoma have metastatic or advanced disease. The Committee noted that the company then assumed 43% of people have oncology treatment (based on data from the 2013 National Oesophago-Gastric Cancer Audit), representing those who have palliative oncology treatment, with the remaining 57% having best supportive care, palliative surgery, and endoscopic or radiological palliation. Of this 43%, the company had then used a figure of 77% (also based on the audit data) to calculate the number of people who would have chemotherapy, with the remaining 23% having either radiotherapy or chemoradiotherapy. The Committee considered the audit data to be a reliable source. It noted that, in the final step of the company's population estimate, the company assumed 30% of people who have chemotherapy go on to have second‑line chemotherapy. The Committee heard from the ERG that this was based on a survey of UK treatment patterns that the company had conducted 1 month before COUGAR II had been published, and that since then the positive results for docetaxel in the COUGAR II study may have led to increased use of chemotherapy in general. In the absence of other data, the Committee considered the value of 30% was reasonable, but believed there to be some uncertainty because of changes to clinical practice since the results of COUGAR II were published. The Committee estimated that based on a population size of 9198 people with gastric cancer or gastro–oesophageal junction adenocarcinoma, and using all other company assumptions, the total population would be around 731 people. The Committee considered that the total population size for the ramucirumab marketing authorisation for advanced gastric cancer or gastro–oesophageal junction adenocarcinoma, for people who have already had platinum therapy, was likely to be greater than 731 people, but fewer than 1,000 people. The Committee, noting that this range was considerably less than 7,000, concluded that the small population size criterion was met.

The Committee concluded that the end‑of‑life considerations could not be applied for ramucirumab in combination with paclitaxel or as a monotherapy, because in both cases the extension‑to‑life criterion of at least an additional 3 months was not met.

The Committee discussed how innovative ramucirumab is in its potential to make a significant and substantial impact on health-related benefits. It noted that ramucirumab is the first biologic agent to have shown efficacy in people whose disease had progressed after chemotherapy, and that it provides an active treatment option for people for whom cytotoxic chemotherapy is not appropriate. Mindful of its conclusion in section 4.1 about the prognosis for people with this disease, the Committee also agreed that this is an area of high unmet medical need, with consequences not only for people but also carers and family members. However, the Committee concluded that all health‑related benefits had been adequately captured by the QALYs in the model, and it agreed that ramucirumab did not offer a step change in the treatment of this disease.

The Committee concluded that ramucirumab in combination with paclitaxel for the treatment of adults with advanced gastric cancer or gastro–oesophageal junction adenocarcinoma, with disease progression after platinum and fluoropyrimidine chemotherapy, was not a cost‑effective use of NHS resources at the usual range of ICERs (£20,000 to £30,000 per QALY). It further concluded that ramucirumab monotherapy for the treatment of adults with advanced gastric cancer or gastro–oesophageal junction adenocarcinoma with disease progression after platinum or fluoropyrimidine chemotherapy, for whom treatment in combination with paclitaxel is not appropriate was not a cost‑effective use of NHS resources at the usual range of ICERs (£20,000 to £30,000 per QALY).

The Committee discussed whether there were any equality issues it should consider before making its recommendations. It noted the company submission and comments received during consultation had stated that the lack of an available licensed treatment after disease progression on chemotherapy can lead to inequalities in access in different parts of England. It considered this was an issue of geographical variation and it was not aware that the potential inequality in access applied to any protected groups covered by the equality legislation. It also considered that any NICE recommendation would be applied consistently across England, thereby reducing variation in practice. It concluded that there was no unfairness or unlawful discrimination, and as a result there were no equality issues, and it did not need to alter its recommendations in any way.

The Committee considered whether it should take into account the consequences of PPRS 2014, and in particular the PPRS payment mechanism, when appraising ramucirumab. The Committee noted NICE's position statement in this regard, and 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 on the relevance of the PPRS to this appraisal. It therefore concluded that the PPRS payment mechanism was not applicable when considering the cost effectiveness of ramucirumab.