The Appraisal Committee reviewed the data available on the clinical and cost effectiveness of pemetrexed, having considered evidence on the nature of NSCLC and the value placed on the benefits of pemetrexed by people with the condition, those who represent them, and clinical specialists. It also took into account the effective use of NHS resources.
The Committee discussed current UK clinical practice for the treatment of NSCLC. It noted that the manufacturer had limited its analysis to comparisons with gemcitabine/cisplatin, gemcitabine/carboplatin and docetaxel/cisplatin. The Committee heard from clinical specialists that current UK clinical practice was to combine gemcitabine with a platinum drug (usually cisplatin) in the majority of cases. It also heard that there were still some centres that used carboplatin as they could not administer cisplatin because of the hydration required, and possibly because some consider carboplatin to be less toxic.
The Committee discussed the additional comparator presented by the manufacturer (docetaxel/cisplatin) and the ERG's concern that vinorelbine had been excluded. The Committee heard from clinical specialists that docetaxel and vinorelbine are not widely used in the UK because of their adverse-event profiles, in particular the higher rates of febrile neutropenia compared with those seen with pemetrexed and gemcitabine. However, the Committee heard from clinical specialists that docetaxel requires fewer hospital visits than gemcitabine, and so it is occasionally used in areas where patients have difficulty getting to hospital. The Committee noted market research data presented by the manufacturer that confirmed that gemcitabine was the main treatment regimen used in the UK, with an 85% market share. Vinorelbine was in second place with an 11% market share. The Committee heard from clinical specialists that the 11% market share of vinorelbine could be an overestimate because it could include use in other indications. The Committee concluded that the gemcitabine/cisplatin combination was the principle comparator in UK clinical practice for the first-line treatment of NSCLC.
The Committee considered the evidence on the clinical effectiveness of pemetrexed/cisplatin compared with gemcitabine/cisplatin. It noted that the JMDB trial was well conducted and considered its results to be robust. The Committee heard from the clinical specialists that the histological subtyping was an important factor in predicting response to pemetrexed. It also heard that the improved overall survival with pemetrexed/cisplatin seen in the JMDB trial in the adenocarcinoma and large-cell carcinoma subgroups has been replicated in other studies. Additionally the Committee noted that pemetrexed had not been proven to be effective in the non-specified histology subgroup. It was mindful that the p value for interaction (see 3.13) supported the hypothesis that the differences between the subgroups was real and not due to chance. The Committee concluded that there is evidence to support a true difference in response to pemetrexed between histological subtypes, although the pathophysiological basis for this is not known.
The Committee then discussed whether the results of the JMDB trial were generalisable to UK clinical practice, with particular reference to routine identification of histological subtypes and numbers of treatment cycles recommended. It heard from clinical specialists that histological identification of patients with non-squamous disease to determine whether they have adenocarcinoma or large-cell carcinoma was not common practice in the UK. However the Committee was satisfied that there would not be problems with doing this in practice because pathology services across the UK can perform such histological diagnoses.
The Committee noted that 4 cycles of chemotherapy was considered standard UK clinical practice, whereas the JMDB trial had allowed up to 6, with an average of 4.4 actually being administered. The clinical specialists stated that a reduction in the number of cycles from 4.4 to 4 was unlikely to affect the clinical outcomes of the trial. The Committee concluded that pemetrexed/cisplatin was more clinically effective than gemcitabine/cisplatin in patients with adenocarcinoma and large-cell carcinoma.
The Committee considered the indirect comparison of pemetrexed/cisplatin with gemcitabine/carboplatin and docetaxel/cisplatin. It noted the manufacturer's exclusion of comparators such as vinorelbine. It considered that even though its use in the UK was low, the omission was inappropriate because it excluded additional information and data from the analysis. The Committee was also mindful of the concerns of the ERG over the methodology used by the manufacturer, and of the fact that the indirect comparisons presented in the manufacturer's submission were potentially flawed because of the exclusion of relevant comparators and the chosen statistical method. However, the Committee noted that the gemcitabine/cisplatin combination was the principle comparator in UK clinical practice for the first-line treatment of NSCLC. It also noted evidence from the clinical specialists and patient expert that suggested that gemcitabine/cisplatin was as effective or more effective than gemcitabine/carboplatin or docetaxel/cisplatin. The Committee concluded that its concerns about the indirect comparison did not prevent it from concluding that pemetrexed/cisplatin is clinically effective in UK clinical practice.
The Committee heard from the patient expert and clinical specialists that pemetrexed was valued by patients because of its favourable adverse-event profile, in particular the lower incidences of febrile neutropenia and alopecia. In addition, patients preferred pemetrexed's shorter infusion time and the fewer hospital visits needed for treatment compared with gemcitabine. The Committee concluded that the increased survival in the adenocarcinoma and large-cell carcinoma subpopulations and lower toxicity demonstrated in the JMDB trial for pemetrexed/cisplatin was clinically significant when compared with gemcitabine/cisplatin, especially when taking into account the overall low survival rates for NSCLC.
The Committee considered the manufacturer's original cost-effectiveness analysis and the ERG's critique. The Committee noted that the original model did not replicate the results of the JMDB trial, especially with respect to the three primary clinical outcomes (overall survival, progression-free survival and response rate). The Committee agreed with the ERG that the model should be able to reproduce the JMDB trial results, because the JMDB trial data are the primary source of clinical data used in the model. The Committee also noted the other problems identified by the ERG and was concerned that the submitted model had not been adequately quality assured. The Committee concluded that on the basis of the evidence presented, the cost effectiveness of pemetrexed/cisplatin had not been proven despite the apparently favourable ICERs in the manufacturer's original submission.
The Committee subsequently considered the revised analysis submitted by the manufacturer. The Committee considered that reducing the number of cycles to four and therefore diverging from the trial was inappropriate for a trial-based analysis. It also considered that the utility values used for progressive states were not appropriate. The Committee concluded that the ERG's exploratory analysis of the manufacturer's revised analysis produced the most plausible estimates. The Committee noted that the ERG's exploratory analysis resulted in ICERs above £48,000 per QALY gained and therefore suggested that pemetrexed/cisplatin was not cost effective. However, the Committee considered that because this analysis only covered the duration of the trial it was inappropriate to conclude cost ineffectiveness from this, although it provided useful additional validation for the subsequently revised Markov model, and the ERG analyses of that.
The Committee considered the manufacturer's modified Markov model and ERG comments on it. The Committee was concerned that some issues of face validity identified by the ERG had not been appropriately addressed. The Committee noted that although reducing the average number of cycles from 4.4 to 4 did not affect the conclusion that pemetrexed was clinically effective, setting a maximum of 4 cycles would affect the conclusions of the cost-effectiveness analysis. It considered that the manufacturer should have taken some account of the probable lower effectiveness. The Committee noted the new errors identified by the ERG that suggested the new analysis had not been sufficiently quality assured. The Committee concluded that the submitted modified Markov model was still not suitable for drawing conclusions because of its inability to replicate the trial results accurately and the lack of quality assurance.
The Committee considered the ERG's exploratory analyses based on the manufacturer's modified Markov model. It was mindful that there were limitations with the data available and that the analyses did not consider the inherent uncertainty in the point estimates through probabilistic sensitivity analysis. The Committee noted that the ERG's estimates of survival were based on individual patient data and that they adequately represented the trial results, in particular the long-term extrapolation. The Committee concluded that the ERG's exploratory analyses were sufficiently robust to allow conclusions to be drawn about the cost effectiveness of pemetrexed/cisplatin.
The Committee noted that the ICERs estimated by the ERG's exploratory analysis were all under £30,000 per QALY gained regardless of the population examined for six cycles of chemotherapy. The Committee noted that when the number of cycles was reduced to four and the ERG's calculations for reduced effectiveness were included, the ICERs were between £20,000 and £30,000 per QALY gained for non-squamous NSCLC and between £17,000 and £25,000 per QALY gained for adenocarcinoma and large cell carcinoma. The Committee therefore concluded that pemetrexed/cisplatin was a cost-effective use of NHS resources based on the evidence available.
The Committee acknowledged that generic versions of gemcitabine have recently become available and that the price was currently subject to change. It noted the ERG's view that when including any substantial price reduction for gemcitabine in the model, pemetrexed/cisplatin was no longer cost-effective compared with gemcitabine/cisplatin. However, it also noted that there was no nationally available price for the generic versions, and that local prices were likely to vary considerably. The Committee concluded that, since the published list price for gemcitabine had not changed, the cost-effectiveness analysis on which it had to base its decision was that described in section 4.13. The Committee considered that the guidance for pemetrexed should be reviewed early if there is a substantial change to the nationally available price of gemcitabine in the NHS.
The Committee considered that current UK clinical practice was to use up to four cycles of gemcitabine/cisplatin as first line-chemotherapy for the treatment of NSCLC. Consequently the Committee considered that the clinical-effectiveness evidence from the JMDB trial, the clinical specialists and patient expert was sufficient and robust enough to demonstrate the clinical effectiveness of pemetrexed/cisplatin in patients with adenocarcinoma and large-cell carcinoma. The Committee noted that pemetrexed/cisplatin had not been shown to be any more effective than gemcitabine/cisplatin in patients with non-squamous NSCLC with unspecified histology. The Committee considered that the ERG's exploratory analysis had demonstrated that the ICERs for pemetrexed/cisplatin were between £17,000 and £25,000 per QALY for adenocarcinoma or large-cell carcinoma. It therefore recommended pemetrexed as an option for the first-line treatment of patients with adenocarcinoma or large-cell carcinoma. The Committee considered that this guidance should be reviewed early if there is any significant change in the price of generic gemcitabine.