Everolimus for advanced renal cell carcinoma after previous treatment

The committee heard from clinical and patient experts that there are limited treatment options for people with advanced RCC.

The committee heard from the clinical and patient experts that advanced RCC is a relatively rare cancer, and noted the views of experts on the severity of the disease. The committee also heard that people having second-line chemotherapy valued the increased life expectancy offered and were prepared to cope with the adverse effects of these treatments. The committee noted the increased frequency of adverse events (including serious adverse events) associated with everolimus treatment in the RECORD‑1 trial. In particular, the committee noted that the most common grade 3 or 4 adverse events suspected to be related to everolimus treatment were anaemia, hyperglycaemia, stomatitis, fatigue, hypercholesterolaemia and dyspnoea. However, the committee was advised by the clinical and patient experts that everolimus would be tolerated by most people with advanced RCC, and that people having everolimus would do so after having had sunitinib as a first-line treatment, and so would be prepared for the adverse effects associated with everolimus. The committee discussed the risk of pneumonitis and immunosuppression associated with everolimus. The clinical expert confirmed that, although pneumonitis and immunosuppression had been associated with everolimus in clinical practice, these adverse events would stop on stopping treatment and were therefore considered manageable.

The committee discussed the evidence of clinical effectiveness (from the RECORD‑1 trial) of everolimus in people with advanced RCC whose disease had progressed during, or within 6 months of stopping, vascular endothelial growth factor targeted treatment. The committee noted that most of the trial population had a good performance status. The clinical expert highlighted that, in clinical practice, only people with a good performance status would be considered for second-line therapy because people with a poorer performance status would be too ill to have any active treatment. Therefore, the committee accepted that the trial population was likely to be similar to people considered for second-line therapy in UK clinical practice. The committee also agreed that the RECORD‑1 trial was of good methodological quality and therefore the results could be considered robust.

The committee discussed the results of the RECORD‑1 placebo-controlled trial. The committee agreed that the results showed that everolimus plus best supportive care had increased progression-free survival by approximately 3 months compared with placebo plus best supportive care. The committee acknowledged that the relative estimates of overall survival according to the intention-to-treat analyses were biased because 81% of people had crossed over to have everolimus in the trial. The committee heard from the clinical expert that an increase in progression-free survival would be expected to result in an increase in overall survival because gains in overall survival had been seen in clinical practice with the introduction of sequential chemotherapy for advanced RCC. The committee noted the meta-analysis submitted by the manufacturer and accepted that a 1.4‑month increase in overall survival per 1‑month increase in progression-free survival for patients with advanced RCC who had had prior therapy was plausible.

The committee agreed that it was appropriate to adjust the intention-to-treat results (which gave a median overall survival estimate of 14.8 months for everolimus plus best supportive care and 14.4 months for best supportive care alone) to control for the crossover using statistical modelling techniques. However, the committee agreed that any estimate of overall survival obtained using statistical modelling would be subject to uncertainty.

The committee acknowledged that the manufacturer had updated both the Inverse Probability of Censoring Weight (IPCW) and the Rank Preserving Structural Failure Time (RPSFT) analyses in response to comments received during consultation. The committee noted that the resulting estimates of overall survival were 16.2 and 16.1 months with everolimus plus best supportive care and 9.6 and 7.9 months with best supportive care using the IPCW and RPSFT methods respectively. The differences in overall survival were 6.5 months and 8.2 months respectively. The evidence review group (ERG) conducted exploratory analyses of the manufacturer's estimates derived using the RPSFT method and noted that the estimates of overall survival were 14.1 months with everolimus plus best supportive care and 8.9 months with best supportive care (difference in overall survival of 5.2 months). The committee noted that the overall survival estimates for both everolimus and best supportive care were higher with the ERG's exploratory analyses than the manufacturer's analyses. The committee concluded that, although there was sufficient evidence that everolimus increased progression-free and overall survival compared with best supportive care, the exact magnitude of the overall survival gain was uncertain because it was based on modelled data as opposed to data directly seen in the trial. However, the committee accepted that overall survival gain would be more than 3 months.

The committee noted that the key factor in determining the cost effectiveness was the estimate of overall survival and discussed the IPCW and the RPSFT methods used to estimate this from the RECORD‑1 trial data. It heard from the ERG that it considered the RPSFT method to be more methodologically robust because the IPCW method assumes there are no unmeasured confounders. In addition, the committee understood that the manufacturer's revised IPCW analysis contained a number of unexplained differences between the original and revised models, and so the ERG could not conduct a full critique of the revised IPCW analysis. The committee also noted that the RPSFT method had been used previously in NICE's technology appraisal guidance on sunitinib for the treatment of gastrointestinal stromal tumours. The committee therefore concluded that, in this instance, it was more appropriate to evaluate the cost effectiveness of everolimus based on the estimates generated using the RPSFT method.

The committee discussed the validity of the estimates of overall survival from the manufacturer's and ERG's RPSFT analyses. The committee noted the ERG's criticism that the manufacturer's extrapolation of long-term survival in the best supportive care arm was still not based on all of the available data (it was based on the mean of cycles 5 and 6 derived from the RPSFT analysis) and that these data may not be representative of the whole trial population. The committee accepted that the use of a Weibull distribution was a more appropriate method for fitting and extrapolating the curve because all available data were used. The committee therefore agreed that this method produced the most plausible estimate of overall survival.

The committee accepted for this appraisal that the costs and utilities associated with living in the 'progressed disease' health state were similar in patients having everolimus and patients having best supportive care. It also agreed that the incremental difference in overall survival was a key factor in determining the cost effectiveness. The committee acknowledged comments received that overall survival with best supportive care in the ERG's exploratory analyses using the Weibull distribution (8.9 months) was higher that was seen in clinical practice, and that the estimate in the manufacturer's analysis (7.9 months) was more likely to reflect clinical practice. The committee noted that the difference in overall survival between patients having everolimus and those having best supportive care was 8.2 months in the manufacturer's revised RPSFT analysis and 5.2 months in the ERG's revised RPSFT analysis. It noted the earlier conclusion that an increase in overall survival of 1.4 months per 1 month of increased progression-free survival was plausible. Therefore, the committee agreed that the incremental overall survival derived using the manufacturer's revised RPSFT analysis (8.2 months) was greater than expected, based on the increase in progression-free survival of 3 months seen in the RECORD‑1 trial. The committee accepted that the ERG's estimate of overall survival for patients having best supportive care using the RPSFT analysis was higher than seen in clinical practice, but the incremental difference in overall survival for everolimus versus best supportive care (5.2 months) was more plausible than that derived by the manufacturer and was based on all of the available data.

The committee then discussed the manufacturer's updated estimate of cost effectiveness derived using the RPSFT analysis. The revised deterministic base-case analysis resulted in an incremental cost-effectiveness ratio (ICER) for everolimus plus best supportive care compared with best supportive care alone of £49,300 per quality-adjusted life year (QALY) gained. The committee understood that the updated estimate also included a revised patient access scheme that had been agreed with the Department of Health. The committee then discussed the results of the manufacturer's probabilistic sensitivity analysis using the adjusted 95% confidence interval around the hazard ratio for overall survival, which gave a mean ICER of £49,500 per QALY gained. The committee noted that this analysis incorporated confidence intervals for the hazard ratio for overall survival adjusted by the manufacturer, rather than the limits as derived directly from the RPSFT analysis. The committee noted that the lower limit of the 95% confidence interval for the hazard ratio for overall survival (0.27) had been derived from clinical opinion data collected by the manufacturer. The committee noted that these data were from a small sample of clinicians and details about the distribution of values within the dataset had not been provided. The committee therefore agreed that these data were likely to be biased. The committee therefore agreed that it would not consider further the results of this analysis.

The committee discussed the ERG's critique of the manufacturer's probabilistic and one-way sensitivity analyses and accepted that the ERG's criticisms of these analyses were valid. The committee noted that the ERG's re-run of the probabilistic sensitivity analysis, which incorporated the 95% confidence interval obtained from the RPSFT analysis, resulted in a mean ICER for everolimus plus best supportive care compared with best supportive care alone of £51,700 per QALY gained. This gave a 24.0% and 52.7% probability of everolimus plus best supportive care being cost effective compared with best supportive care alone if the maximum acceptable amount to pay for an additional QALY gained was £30,000 or £50,000 respectively. The committee concluded that, because of the errors identified in the manufacturer's analysis, the ERG's probabilistic analysis was the most plausible.

The committee then discussed other aspects of the manufacturer's model and the critique by the ERG. The committee considered that the time horizon and discounting in the analyses were appropriate. However, the committee had concerns about the validity of some of the assumptions used in the economic model. Firstly, it noted that all patients entered the economic model in the 'stable disease without adverse events' health state. The committee heard from the clinical expert that, in practice, patients eligible for treatment would present with progressed disease and it was likely that some people starting a second-line therapy for advanced RCC experienced adverse events. Secondly, the committee was concerned about the model assumption that the costs of managing associated adverse events would apply for only 1 treatment cycle. However, the committee heard from the clinical expert that adverse events would be managed by 'drug holidays' or dose reduction and therefore treatment of adverse events would not be expected to incur significant ongoing costs. The committee also heard from the clinical expert that the primary ongoing adverse event with everolimus was fatigue, but that this was common to all cancer treatments and there were currently no treatments for its management. Therefore, the committee agreed that the cost estimates used for adverse events in the model were acceptable.

The committee noted that the utility estimates in the model were neither directly obtained nor mapped from the RECORD‑1 trial. The committee noted that the estimates of utility for each of the disease states were similar. The committee accepted that a larger decrement in utility may be plausible when a person moves from a 'stable disease' health state to a 'progressed disease' health state. The committee noted comments from the ERG and the results of the one-way sensitivity analyses, which showed changes in utility estimates had little effect on the ICERs. The committee agreed that, although the utility estimates were subject to some uncertainty, the utility assumptions in the economic model were acceptable.

The committee was aware of the 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 which 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 patients 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 the assumptions used in the reference case economic modelling are plausible, objective and robust.

The committee then discussed whether everolimus as a second-line treatment for advanced RCC fulfilled the criteria for consideration as a life-extending, end-of-life treatment. It was aware that in England and Wales the total number of people who would be eligible for treatment with everolimus was less than 4,000. The committee heard from the clinical expert that the life expectancy for people with advanced RCC having best supportive care alone was unlikely to be greater than 24 months and was potentially as low as 5 months. The committee also noted that the evidence from the RPSFT analysis suggested that everolimus increased survival by more than 3 months compared with best supportive care. In summary, the committee was satisfied that everolimus met the criteria for being a life-extending, end-of-life treatment, and that the evidence presented for this consideration was sufficiently robust.

The committee then discussed whether, in view of the estimates of cost effectiveness, everolimus was an appropriate use of NHS resources for a life-extending, end-of-life treatment. The committee considered 2 key issues: first the central estimate of the ICERs, and second the robustness and certainty of the ICER. It noted that the deterministic ICER of £49,300 per QALY gained was high and close to the range considered acceptable for end-of-life treatments. The committee also noted the wide confidence intervals and uncertainty introduced by the novel methodology used to obtain this ICER. Therefore the committee considered the importance of considering the mean probabilistic ICER of £51,700 per QALY gained from the ERG's exploratory probabilistic sensitivity analysis (incorporating the revised patient access scheme). It noted that this ICER was higher than those considered acceptable for end-of-life treatments to date. The committee noted that the ERG's probabilistic sensitivity analysis had indicated that, if the maximum acceptable amount to pay for an additional QALY gained was £30,000, the probability that everolimus was cost effective compared with best supportive care alone was only 24.0%. It also noted that, if the maximum acceptable amount to pay for an additional QALY gained was £50,000, the probability that everolimus was cost effective compared with best supportive care alone was only 52.7%. The committee concluded that, because the ICERs were subject to considerable uncertainty and were high, the magnitude of additional weight that would need to be assigned to the original QALY benefits in this patient group was too high for the cost effectiveness of the drug to fall 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 it could not recommend everolimus for the second-line treatment of advanced RCC as a cost-effective use of NHS resources.

The committee considered whether there were any subgroups of patients for whom everolimus would be considered a cost-effective use of NHS resources, and whether NICE's duties under the equalities legislation required it to alter or to add to its recommendations in any way. The committee noted that no subgroups of patients had been identified and agreed that that there are no specific equality issues relevant to this appraisal.

This appraisal was a Cancer Drugs Fund reconsideration of the published NICE technology appraisal guidance on everolimus for the second-line treatment of advanced RCC. Everolimus has been available through the Cancer Drugs Fund. In its revised submission updating its cost-effectiveness analysis, the company:

• introduced a revised patient access scheme that provides a simple discount to the list price of everolimus (the level of the discount is commercial in confidence)

• updated unit cost data to 2016 values and

• presented evidence comparing everolimus with axitinib, based on the updated NICE scope for the Cancer Drugs Fund reconsideration.