Axitinib for treating advanced renal cell carcinoma after failure of prior systemic treatment

4.4 The Committee examined the clinical evidence from the AXIS trial, which compared axitinib with sorafenib. The Committee noted that the trial was well conducted and the relevant outcomes were assessed in line with the scope of the appraisal. However, it noted the difficulties in interpreting the AXIS trial results in this appraisal because of the lack of comparisons with any of the scope comparators. The Committee noted that the better progression‑free survival results for axitinib (6.7 months for the axitinib group compared with 4.7 months for the sorafenib group [HR 0.67, 95% CI 0.54 to 0.81, p<0.0001]) did not translate into statistically significant overall survival benefits (20.1 months for the axitinib group compared with 19.2 months for the sorafenib group [HR 0.97, 95% CI 0.80 to 1.17, p=0.37]) for the full trial population. The Committee heard the company's explanation that this could be a result of the use of subsequent cancer treatments after progression. It was satisfied with the health‑related quality‑of‑life data collected and assessed in the AXIS trial using both generic and disease‑specific instruments. The Committee concluded that AXIS was a well‑conducted trial, which showed that axitinib provided clinical benefit to people who have been treated previously with sunitinib or a cytokine. Because there was no relevant comparator in the AXIS trial, the Committee concluded that its discussion of the efficacy of axitinib would need to be based on the results of the indirect and the simulated treatment comparisons performed by the company. It also concluded that it was reasonable to separate out the results for patients who had received only prior cytokines from people who had received prior tyrosine kinase inhibitors.

4.5 The Committee heard from the clinical experts and the company that approximately 1% of patients would receive only prior cytokines. Acknowledging the clinical experts' views (see section 4.2), the Committee noted that the prior‑cytokine population has been diminishing since the introduction of sunitinib and pazopanib and NICE's approval of these as first‑line treatments. It considered that 'prior‑cytokine patients' would, in practice, be given sunitinib or pazopanib, despite previous treatment with cytokines. The Committee examined the analyses performed to generate treatment comparisons of axitinib with best supportive care, sunitinib and pazopanib in the prior‑cytokine group.

4.6 The Committee discussed the analysis performed to generate a comparison of axitinib with best supportive care in the prior‑cytokine group. It noted that the evidence for the indirect comparison was based on the AXIS trial (which compared axitinib with sorafenib) and the TARGET trial (which compared sorafenib with placebo). The Committee accepted the company's use of placebo as a proxy for best supportive care in the indirect and simulated treatment comparisons. It was aware that patient baseline characteristics were not presented separately for the prior‑cytokine subgroups in the 2 trials. It noted that the 2 trials were not fully comparable in terms of Memorial Sloan‑Kettering Cancer Centre (MSKCC) scores, prior treatments and number of metastatic sites reported. The Committee noted that crossover in the TARGET trial was adjusted by censoring the patients who crossed over and considered that this could have resulted in bias and ultimately affected the robustness of the results of the indirect comparison (progression‑free survival of 11 months for the axitinib group compared with 3.5 months for the best supportive care group, a 7.5‑month difference [HR 0.25, 95% CrI 0.17 to 0.38] and overall survival of 33.5 months for the axitinib group compared with 23.5 months for the best supportive care group, a 10‑month difference [HR 0.63, 95% CrI 0.41 to 0.99]). It also noted that an assumption of proportional hazards, which assumes a constant treatment effect over a lifetime, had been used to derive the survival estimates and had not been tested. The Committee agreed that, although the results might not be robust, the indirect comparisons were adequately performed. The Committee concluded that axitinib was more clinically effective than best supportive care in the prior‑cytokine population.

4.7 The Committee considered the naive comparison performed to compare axitinib with sunitinib in the prior‑cytokine population. It noted that the little evidence available was from single‑arm trials, and that the Evidence Review Group (ERG) agreed with the company that an indirect comparison was not appropriate given the lack of data. The Committee noted that axitinib provided a 5.5‑month extension to overall survival compared with sunitinib in the naive comparison. The Committee concluded that, although the naive comparison was the best option given the available evidence, the results were not robust and were subject to uncertainty.

4.8 The Committee discussed the evidence used in the indirect comparison of axitinib with pazopanib in the prior‑cytokine subgroup, which was based on the AXIS trial (axitinib compared with sorafenib), the TARGET trial (sorafenib compared with placebo) and the VEG105192 trial (pazopanib compared with placebo). The Committee noted that, as a result of the crossover between treatment arms in the VEG105192 trial, the company had only performed an indirect comparison using progression‑free survival as an end point and had performed a naive comparison for the overall survival data. The Committee noted that the progression‑free survival comparison favoured axitinib, and that the naive overall survival comparison suggested that axitinib provided an additional 6.7 months overall survival benefit compared with pazopanib (see section 3.41). The comparison also suggested that best supportive care provided an additional 1.3 months overall survival over pazopanib (24 months for best supportive care compared with 22.7 months for pazopanib, see section 3.42). The Committee heard from the clinical experts that it was not clinically plausible that best supportive care would have a median overall survival of 24 months, or that best supportive care would have a survival benefit over sunitinib or pazopanib. The Committee then considered the exploratory indirect comparison carried out by the ERG for overall survival. It noted that the hazard ratios for overall survival varied from 0.77 (suggesting that axitinib is better than pazopanib) to 1.21 (suggesting that pazopanib is better than axitinib) depending on the approach used to adjust for crossover and that all the hazard ratios were associated with wide confidence intervals (see section 3.56). The Committee agreed that, although evidence in favour of a relative advantage for axitinib was available from the progression‑free survival comparison, the ERG's indirect comparisons for overall survival data were more appropriate than the company's naive comparison, and these did not favour axitinib. The Committee concluded that the results generated from the naive and indirect comparisons for overall survival and progression‑free survival were all subject to substantial uncertainty. The clinical experts agreed with this and did not consider axitinib to be different in effectiveness from sunitinib or pazopanib after prior‑cytokine treatment. The Committee concluded that axitinib was likely to have clinical effectiveness comparable to pazopanib and sunitinib.

4.9 The Committee considered the simulated treatment comparison of axitinib with best supportive care (in line with the NICE scope) performed for the prior‑sunitinib subgroup using evidence from the AXIS trial (which compared axitinib with sorafenib) and the RECORD‑1 trial (which compared everolimus with placebo). It noted that this method of comparison was used to create an adjusted indirect comparison of axitinib with best supportive care in the prior‑sunitinib group. The Committee discussed whether the use of the simulated treatment comparison method could be considered reliable and valid given that it is a relatively new method of treatment comparison. It noted the ERG's comment that it was based on a comparison of 2 single treatment arms without random allocations to treatment. The Committee was aware that crossover also occurred in the RECORD‑1 trial and, although this was adjusted for using the rank‑preserving structural failure time (RPSFT) method, it noted that this method may not be appropriate for subgroups because it assumes the same treatment effect applies across the whole trial population, as well as there being no unobserved factors that should have been controlled for in the analysis. It was also aware that there were key differences between the trial populations in RECORD‑1 and AXIS, which could bias the results of the simulated treatment comparison, such as:

4.10 The Committee examined the plausibility of a post‑progression survival gain with axitinib in the context of the progression‑free and overall survival relationship presented by the company (see section 3.26). It compared the relationship between progression‑free survival and overall survival gain estimated from the company's simulated treatment comparison (1 to 1.6) and that originally modelled by the ERG (1 to 1). It noted that the relationship had been weakened by the inclusion of more studies in which crossover occurred in the updated meta‑analysis (1 to 1.04 for the subgroup of trials with prior treatment on the basis of updated RECORD‑1 analyses), but that this analysis was not robust because it did not properly take into account crossover between treatment arms (the relationship rose to 1 to 1.29 when adjustments for crossover were made). The Committee also heard from the company that active targeted treatments are associated with higher response rates and tumour shrinkage compared with best supportive care, based on Grunwald et al. The Committee agreed that the evidence from Grunwald et al. gave some support to the company's simulated treatment comparison results on the post‑progression survival benefits of axitinib over best supportive care.

4.11 The Committee considered the adverse event profile associated with axitinib that was observed in the AXIS trial. The Committee noted that diarrhoea, which was the most common adverse event, occurred with similar frequency in the axitinib and sorafenib groups. It was aware that hypertension, dysphonia, nausea and hypothyroidism occurred more frequently in the axitinib group, although hand–foot syndrome, rash and alopecia occurred more frequently in the sorafenib group. The Committee also noted the comment from the clinical experts that axitinib was a well‑tolerated drug except for the high occurrence of hypertension, which is common with all tyrosine kinase inhibitors. The patient experts commented that people would be willing to accept these adverse events, and so the Committee concluded that axitinib has a manageable adverse event profile compared with other treatments for advanced renal cell carcinoma.

4.12 The Committee considered the company's economic model and the ERG's critique of the model. It was satisfied that the outlined economic analysis was acceptable for assessing the cost effectiveness of axitinib. The Committee concluded that the appropriate populations and comparators for the economic evaluation had been captured in the model.

4.13 The Committee discussed the assumptions made by the company in developing the economic model. It noted that when alternative survival distributions for progression‑free and overall survival were tested in the scenario analysis, they resulted in sizeable changes to the base‑case result for the prior‑cytokine population and moderate changes for the prior‑sunitinib population (see sections 3.30, 3.34 and 3.54). The Committee concluded that the model results were highly sensitive to the distributions used to extrapolate survival.

4.14 The Committee discussed the plausibility of the survival gains estimated for the prior‑cytokine group from the economic model. It heard from the clinical experts and the patient expert that the overall survival gain of approximately 24 months in the best supportive care group of the prior‑cytokine group is not seen clinically. It noted the company's comment that the implausibility observed may have resulted from the overall survival of 14 months in the placebo arm of TARGET, which was not properly adjusted for crossover. The Committee considered that this possible over‑estimation of the overall survival in TARGET was carried over into the overall survival results in the indirect comparison and ultimately affected the model results for the best supportive care group. The Committee heard from both the clinical experts and the company that the lower 95%confidence interval estimate of 17.46 months for overall survival in the best supportive care group was more clinically plausible than the median of 24 months. The Committee was uncertain whether pazopanib and sunitinib were extendedly dominated by axitinib (that is, a quality‑adjusted life year [QALY] is attained at a higher cost with pazopanib or sunitinib than with axitinib because the incremental cost effectiveness ratio (ICER) for pazopanib or sunitinib compared with best supportive care is higher than that for axitinib compared with best supportive care), as in the company's naive economic analysis (see sections 3.42 and 3.57). It also noted that the ICERs were sensitive to some of the parameters and assumptions used in the model (such as the utility values, the value of the survival parameters and the type of distribution used to extrapolate survival). The Committee concluded that the company's base‑case ICER of approximately £55,300 per QALY gained (with the patient access scheme applied) may have been over‑estimated based on the unlikely overall survival gains with best supportive care in the prior‑cytokine population, and that there was evidence that the true ICER for the comparison of axitinib with best supportive care in the prior‑cytokine population could be lower if the true overall survival for best supportive care was closer to 17.46 months.

4.15 The Committee considered the uncertainty around the company's base‑case estimates of £33,500 and £52,900 per QALY gained in the prior‑sunitinib group. It recognised that the use of the simulated treatment comparison method to derive a best supportive care comparison for axitinib in the group was the greatest source of uncertainty. The Committee discussed the plausibility of the survival gains estimated for the prior‑sunitinib group (figures are commercial in confidence; see section 3.16). It noted that the median survival gain difference between axitinib and best supportive care estimated directly from the trials was increased by 63% when modelled (median overall survival estimated for the placebo group was 8.3 months compared with 15.2 months for axitinib). Furthermore, the Committee noted that an implausibly high proportion of the total QALY gains with axitinib (compared with best supportive care) in the prior‑sunitinib group was observed after progression when active treatment with axitinib had stopped. It noted that this was not a feature in either the prior‑cytokine analysis or the AXIS trial results. The Committee examined the exploratory analysis performed by the ERG and company in which it was assumed that there was no QALY difference between axitinib and best supportive care in the prior‑sunitinib group after progression. When it was also assumed that there was no cost difference after progression the resultant ICER was approximately £52,900 per QALY gained (with the patient access scheme applied). The Committee concluded that the results from the simulated treatment comparison and the post‑progression model outputs for the prior‑sunitinib group should be interpreted with caution because they lacked clinical plausibility.

4.16 The Committee also noted that use of the RENCOMP indirect comparison of axitinib and best supportive care gave higher ICER values than the company's base‑case result. When the RENCOMP method was used rather than the simulated treatment comparison, the ICER increased to over £40,000 per QALY gained (with the patient access scheme applied) using the Weibull and Gompertz distributions, suggesting that the method of obtaining a best supportive care comparison was a key driver of the results in this population. The Committee concluded that the results for the prior‑sunitinib group should be interpreted with caution because not all the uncertainties had been fully considered.

4.17 The Committee considered the company's comments and the summary of Grunwald et al. that some evidence exists from metastatic renal cell carcinoma trials showing QALY gains in the post‑progression period in addition to those gained in the progression‑free period when targeted therapies are compared with best supportive care in the prior‑tyrosine kinase inhibitor group (see section 4.10). The Committee noted the evidence from Grunwald et al. suggesting that there is plausible post‑progression benefit because of tumour shrinkage. The Committee concluded that the relationship between progression‑free survival and overall survival gain for the prior‑sunitinib group was likely to lie between the company's estimate and the ERG's estimate, and was therefore likely to be larger than 1.04 and less than 1.6.

4.18 The Committee discussed the available cost‑effectiveness estimates for the prior‑cytokine group. It noted that the ICER for axitinib compared with best supportive care fell below £55,300 per QALY gained with any estimate of the relationship between progression‑free survival and overall survival gain greater than 1 to 1. It noted that the company's base‑case ICER of approximately £55,300 per QALY gained (with the patient access scheme applied) generated for the prior‑cytokine group compared with best supportive care was based on the overall survival for best supportive care of 24 months, but fell to £36,500 per QALY gained if 17.46 months for overall survival for best supportive care was used (see section 4.14). The Committee noted there were uncertainties that might increase or decrease this ICER, and concluded that the most plausible ICER is above the range usually considered to be a cost‑effective use of NHS resources in NICE technology appraisals (between £20,000 and £30,000 per QALY gained). Based on the comparable clinical‑effectiveness evidence in the prior‑cytokine group and the differences in NHS costs, the Committee concluded that axitinib, sunitinib and pazopanib could have comparable cost effectiveness.

4.19 For the prior‑sunitinib population, the Committee noted that, although there was uncertainty in the simulated treatment comparison method, it accepted that the Grunwald et al. analysis was supportive. It considered that a more plausible ICER for the prior‑sunitinib group was likely to lie between the base‑case estimate with a progression‑free to overall survival gain relationship of 1 to 1.6 (approximately £33,500 per QALY gained) and the estimate assuming no survival gain with a survival relationship of 1 to 1 (approximately £52,900 per QALY gained). Given the balance of the evidence, the Committee concluded that the ICER would likely be towards the middle of this range.

4.20 Because the ICERs for both populations were above £30,000 per QALY gained, the Committee discussed whether axitinib for advanced renal cell carcinoma fulfilled the criteria for a life‑extending, end‑of‑life treatment, which are that:

4.21 The Committee agreed that the life expectancy of people with advanced renal cell carcinoma in whom prior cytokines and tyrosine kinase inhibitors such as sunitinib and pazopanib have failed was less than 24 months. It also noted the company's evidence to indicate that axitinib treatment offers an extension to life of at least an additional 3 months in the case of the prior‑sunitinib population, compared with the current NHS treatment of best supportive care (see section 4.9). It noted that the company's estimate of the eligible population for whom axitinib is licensed (most in the prior‑sunitinib group), that is, 1580 people in year 1 and up to 1743 people in year 5, represented a small patient population. The Committee concluded that axitinib was shown to be a life‑extending, end‑of‑life treatment for the prior‑sunitinib population. It also agreed that it was not reasonable to limit this conclusion to people whose prior tyrosine kinase inhibitor was sunitinib, given the growing, if not majority use, of pazopanib as a first‑line treatment. This would leave an unmet need and would not reflect clinical practice. Therefore the Committee concluded that its recommendations should apply to the whole prior‑tyrosine kinase inhibitor population.

4.22 With regard to the end‑of‑life criteria in the prior‑cytokine population, the Committee agreed that the overall survival gains with best supportive care were improbable, and that axitinib was likely to provide a greater than 3‑month survival benefit compared with best supportive care. The Committee considered the comments received from the company of axitinib on the second consultation, on the extension of life provided by axitinib compared with pazopanib being greater than 3 months. The Committee noted that this was supported only by the company's indirect comparison calculations and for progression‑free survival and not for overall survival, for which the ERG had shown the possible superiority of pazopanib (see section 4.8). There was indirect comparison evidence only for axitinib compared with pazopanib, and not with sunitinib. Therefore, the Committee concluded that in the prior‑cytokine population, substantial uncertainty remained around the naive comparisons between axitinib, sunitinib and pazopanib, and the end‑of‑life criteria could only be considered met if best supportive care were the only comparator (see section 4.6).

4.23 The Committee then discussed whether the valuation of the health‑related quality of life necessary for axitinib to be considered a cost‑effective use of NHS resources for the prior‑sunitinib population was reasonable. It discussed both the range of ICER valuations available and the degree of certainty around the estimates. The range of ICERs was £33,500 to £52,900 per QALY gained, and the most plausible valuations depended on the assumptions around the relationship between progression‑free survival and overall survival and were likely to be towards the middle of this range (see sections 4.17, 4.18 and 4.19). The Committee concluded that, although the ICERs were subject to considerable uncertainty and were high, the additional weight from the end‑of‑life criteria that could be assigned to the original QALY benefits in this patient group led to the cost effectiveness of the drug falling within the range currently considered a cost‑effective use of NHS resources. Taking into account both the value of the ICERs and the uncertainty around the ICERs, the Committee concluded that axitinib could be considered a cost‑effective use of NHS resources in the prior‑tyrosine kinase inhibitor population under the supplementary criteria for appraising life‑extending, 'end‑of‑life' treatments.

4.24 The Committee then discussed whether axitinib could be considered a cost‑effective use of NHS resources for the prior‑cytokine population. Although the Committee did not accept the superiority of axitinib over sunitinib and pazopanib and therefore did not accept that the end‑of‑life criteria had been met for this population, it was aware of the comments from the clinical experts that axitinib, sunitinib and pazopanib were used interchangeably in clinical practice (see section 4.3). The Committee took both this and the cost‑effectiveness uncertainties into account. It concluded, first, that axitinib was comparable to the alternative treatments that had been recommended by NICE that meet the 'end‑of‑life' criteria (sunitinib for the first-line treatment of advanced and/or metastatic renal cell carcinoma and pazopanib for the first-line treatment of advanced renal cell carcinoma) and, second, that only a very small population was now included in the prior‑cytokine group and that more uncertainty could be accepted in these circumstances. The Committee therefore concluded that an ICER compared with best supportive care above but close to the upper limit of the normal range was acceptable, and axitinib could be considered a cost‑effective use of NHS resources in the prior‑cytokine population.

4.25 The Committee further considered its recommendation. It noted that the marketing authorisation of axitinib is for treating adults with advanced renal cell carcinoma, after failure of prior treatment with sunitinib or a cytokine and does not include previous treatment with pazopanib. However it reiterated that it was not reasonable to limit its recommendation to people whose prior tyrosine kinase inhibitor was sunitinib, given the growing, if not majority use, of pazopanib as a first‑line treatment. This would leave an unmet need and would not reflect clinical practice. The Committee recognised that the requirement to provide funding by the relevant health bodies (clinical commissioning groups, NHS England and local authorities) within 3 months of its date of final guidance publication applies only within the marketing authorisation (see implementation sections 5.1 to 5.3).

4.26 The Committee noted the comments made by the company and patient organisations regarding the 'innovativeness' of axitinib. They stated that axitinib was expected to offer a step‑change in the second‑line management of advanced renal cell carcinoma by improving survival beyond what is expected with best supportive care, while maintaining health‑related quality of life. The Committee understood this, as well as noting the needs of patients for further treatment options, but considered that there were no additional gains in health‑related quality of life over those already included in the QALY calculations. Therefore, the Committee concluded that the innovative aspects of axitinib with regard to patient benefits were already incorporated in the economic model and analyses.

4.27 The Committee discussed potential equality issues and gave particular consideration to avoid unlawful discrimination against any group of people on the grounds of race, gender, disability, religion or belief, sexual orientation, age, gender reassignment, pregnancy and maternity. The Committee noted the following potential equality issues raised by the patient experts, patient organisations and NHS organisations: