Lung cancer (non-small-cell, anaplastic lymphoma kinase fusion gene, previously treated) - crizotinib: appraisal consultation document

The Department of Health has asked the National Institute for Health and Care Excellence (NICE) to produce guidance on using crizotinib in the NHS in England and Wales. The Appraisal Committee has considered the evidence submitted by the manufacturer and the views of non-manufacturer consultees and commentators, and clinical specialists and patient experts.

This document has been prepared for consultation with the consultees. It summarises the evidence and views that have been considered, and sets out the draft recommendations made by the Committee. NICE invites comments from the consultees and commentators for this appraisal (see appendix B) and the public. This document should be read along with the evidence base (the evaluation report), which is available from www.nice.org.uk.

The Appraisal Committee is interested in receiving comments on the following:

  • Has all of the relevant evidence been taken into account?
  • Are the summaries of clinical and cost effectiveness reasonable interpretations of the evidence?
  • Are the provisional recommendations sound and a suitable basis for guidance to the NHS?
  • Are there any aspects of the recommendations that need particular consideration to ensure we 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?

Note that this document is not NICE's final guidance on this technology. The recommendations in section 1 may change after consultation.

After consultation:

  • The Appraisal Committee will meet again to consider the evidence, this appraisal consultation document and comments from the consultees.
  • At that meeting, the Committee will also consider comments made by people who are not consultees.
  • After considering these comments, the Committee will prepare the final appraisal determination (FAD).
  • Subject to any appeal by consultees, the FAD may be used as the basis for NICE’s guidance on using crizotinib in the NHS in England and Wales.

For further details, see the 'Guide to the technology appraisal process' (available at www.nice.org.uk).

The key dates for this appraisal are:

Closing date for comments: 18 April 2013

Second Appraisal Committee meeting: 30 April 2013

Details of membership of the Appraisal Committee are given in appendix A, and a list of the sources of evidence used in the preparation of this document is given in appendix B.

 

Note that this document is not NICE's final guidance on this technology. The recommendations in section 1 may change after consultation.

1 Appraisal Committee's preliminary recommendations

1.1 Crizotinib is not recommended, within its marketing authorisation, for previously treated anaplastic-lymphoma-kinase-positive advanced non-small-cell lung cancer.

1.2 People currently receiving crizotinib that is not recommended according to 1.1 should be able to continue treatment until they and their clinician consider it appropriate to stop.

2 The technology

2.1 Crizotinib (Xalkori, Pfizer) is a selective small-molecule inhibitor of the anaplastic lymphoma kinase (ALK) receptor tyrosine kinase and its oncogenic variants (that is, ALK fusion events and selected ALK mutations). It has been granted a conditional UK marketing authorisation for treating 'adults with previously treated anaplastic lymphoma kinase (ALK)-positive advanced non-small-cell lung cancer (NSCLC)'.

2.2 The summary of product characteristics lists the following as the most common adverse reactions associated with crizotinib treatment: visual impairment, diarrhoea, nausea, vomiting, constipation, oedema, fatigue, decreased appetite, neutropenia and elevated aminotransferases. For full details of adverse reactions and contraindications, see the summary of product characteristics.

2.3 The acquisition cost of crizotinib is 4689 for 1 pack of 60200 mg (or 250 mg) capsules (30-day supply) (excluding VAT; ‘British National Formulary’ [BNF] edition 64). The summary of product characteristics states that the recommended dose of crizotinib is 250 mg twice daily (500 mg daily) taken continuously. It further states that ‘Treatment should be continued until disease progression or unacceptable toxicity. Prolongation of treatment after objective disease progression in selected patients may be considered on an individual basis but no additional benefit has been demonstrated’. Assuming treatment until disease progression, the cost of a course of treatment would be 37,512 using the median progression-free survival in the study PROFILE 1007 as the number of cycles of treatment (that is, 7.7 months or 8 packs of capsules), or 46,890 using the number of treatment cycles calculated from the duration of progression-free survival in the manufacturer’s economic model (that is, 9.6 months or 10 packs of capsules). Using the median number of crizotinib treatment cycles started in PROFILE 1007 (that is, 10.5 months or 11 packs of capsules) the cost of a course of treatment would be 51,579. Costs may vary in different settings because of negotiated procurement discounts.

3 The manufacturer's submission

The Appraisal Committee (appendix A) considered evidence submitted by the manufacturer of crizotinib and a review of this submission by the Evidence Review Group (ERG; appendix B).

3.1 The manufacturer presented evidence on the clinical effectiveness of crizotinib used within the marketing authorisation and in line with the appraisal scope. The manufacturer explained that a comparison with erlotinib, which was a comparator in the scope, was not considered valid and therefore evidence was not provided for this comparison. The manufacturer justified this based on erlotinib’s action as an epidermal growth factor receptor (EGFR) inhibitor, because EGFR mutation does not tend to occur with anaplastic lymphoma kinase (ALK) translocation. Therefore, the evidence provided in the submission considered comparisons of crizotinib with the other 2 remaining comparators in the scope, namely docetaxel and best supportive care.

3.2 The main evidence came from 1 multicentre, randomised phase III efficacy and safety study in patients with advanced or metastatic previously treated ALK-positive non-small-cell lung cancer (PROFILE 1007). In this study, 347 patients were randomised on a 1:1 basis to receive either crizotinib 250 mg twice daily or chemotherapy (pemetrexed 500 mg/m2 [58%]; or docetaxel 75 mg/m2 [42%] if the patient had already received treatment with pemetrexed earlier in their treatment pathway and met eligibility criteria for liver function and peripheral neuropathy). Treatment was given until disease progression, which was defined using the Response Evaluation Criteria in Solid Tumours (RECIST) of independent radiology review, unacceptable toxicity or withdrawal of consent. Treatment after radiographical progression could be continued for patients in the study at the discretion of the investigator. After disease progression, crossover was allowed in both directions.

3.3 The primary outcome in PROFILE 1007 was progression-free survival (PFS). Secondary outcomes included overall survival, objective response rate, duration of response, disease control rate, medication-related adverse events and health-related quality of life (lung cancer specific symptoms and general health status). The primary outcome was reported for the intention-to-treat population and for a post hoc subgroup analysis in which crizotinib was compared with docetaxel in the chemotherapy group. Overall survival was not reported for the subgroup, the manufacturer statedthat this was because of the immaturity of these data.

3.4 The median age of patients recruited into PROFILE 1007 was 51 years in the crizotinib arm, compared with 49 years in the chemotherapy arm. Most patients had adenocarcinoma (94.2% in the crizotinib arm; 92.0% in the chemotherapy arm) and had never smoked (62.4% in the crizotinib arm; 63.8% in the chemotherapy arm). The Eastern Cooperative Oncology Group (ECOG) performance status for patients in PROFILE 1007 was predominantly 0 (approximately 39%) and 1 (52%); only 9% had performance status 2.

3.5 PFS results from PROFILE 1007 werebased on an interim analysis with a cut-off date of March 2012. Results from the intention-to-treat analysis showed that treatment with crizotinib led to a statistically significant increase in median PFS of 4.7 months compared with chemotherapy (7.7 months in the crizotinib arm [95% confidence interval {CI}6.0 to 8.8] compared with 3.0 months in the chemotherapy arm [95% CI 2.6 to 4.3]; hazard ratio [HR] of progression 0.49; 95% CI 0.37 to 0.64; p<0.0001). In the post hoc subgroup analysis, crizotinib was associated with a statistically significant increase in median PFS of 5.1 months compared with docetaxel (7.7 months in the crizotinib arm [95% CI 6.0 to 8.8] compared with 2.6 months in the docetaxel arm [95% CI presented as academic in confidence]; HR of progression 0.30; 95% CI 0.21 to 0.43; p<0.0001).

3.6 Median overall survival results were based on an interim analysis with a cut-off date of March 2012, at which point 28% of the target events (death) had been observed in the crizotinib arm and 27% in the chemotherapy arm. In this analysis,crizotinib was not found to statistically significantly prolong overall survival compared with chemotherapy (20.3 months in the crizotinib arm compared with 22.8 months in the chemotherapy arm, HR 1.02; 95% CI 0.68 to 1.54; p=0.54). The manufacturer stated that these results were likely to have been significantly affected by crossover in the trial(the percentages of patients who crossed over between the treatment arms were reported as confidential data and cannot be reported here). Overall survival data were not presented for the docetaxel subgroup because of the immaturity of the data. The objective response rate was reported for the full population and the docetaxel subgroup. For the full population, objective response rate for crizotinib was 65.3% (95% CI 57.7% to 72.4%) and 19.5% (95% CI 13.9% to 26.2%) for the chemotherapy group; the objective response rate ratio was 3.4 (95% CI 2.4 to 4.7, p<0.0001). For the docetaxel subgroup, objective response rate for crizotinib was 65.7% (95% CI reported as confidential and cannot be reported here) and 6.9% (95% CI reported as confidential and cannot be reported here) for docetaxel; the relative risk was also reported as confidential. Median duration of follow-up was 12.2 months in PROFILE 1007.

3.7 The manufacturer explored the effect of crossover using 4 different approaches, the rank-preserving structural failure time (RPSFT) method, the inverse probability of treatment weighted (IPTW) method, the inverse probability of treatment and censoring weighted (IPTCW) method and the use of a case-matched hazard ratio from ‘real world data’. In addition, 5 different versions of the IPTCW method, which differed in terms of how missing values were imputed, were explored. The RPSFT, IPTCW and IPTW methods used data from PROFILE 1007. The ‘real world data’ method used a published case-matched analysis in which some crizotinib-treated patients with ALK-positive non-small-cell lung cancer from PROFILE 1001 (a single arm safety study of 149 patients treated with crizotinib) were case-matched to patients with ALK-positive non-small-cell lung cancer who did not have crizotinib (control group) from a retrospective analysis. The hazard ratios for death for crizotinib compared with chemotherapy were 0.83 for the RPSFT method and 0.36 for the real world data method. The remaining hazard ratios for the IPTCW methods were presented as confidential data and therefore cannot be reported in this document. The hazard ratios were applied to the economic model to provide modelled mean survival times for chemotherapy. The various estimates from the crossover analyses were compared with the modelled mean survival of 33 months for crizotinib. For the RPSFT method, the modelled mean survival time for chemotherapy was 26.9 months, giving an absolute gain in overall survival for crizotinib compared with chemotherapy of 5.8 months. For the real world data method, the modelled mean survival time for chemotherapy was 11.3 months, giving an absolute gain in overall survival for crizotinib compared with chemotherapy of 21.7 months. The manufacturer dismissed the real world data result on the basis that it did not appear to be supported by evidence from randomised controlled trials, and instead, chose the IPTCW method that used the last observation carried forward for 56 days as an imputation method for missing data. This crossover analysis was referred to as IPTCW5, and resulted in a modelled mean survival time for chemotherapy of 20.8 months, giving an absolute gain in overall survival for crizotinib compared with chemotherapy of 12.3 months. The overall survival hazard ratio resulting from this method was reported as confidential and cannot be reported here.

3.8 Non-comparative evidence for crizotinib was presented from the ongoing, single-arm studies PROFILE 1001 (n=149) and PROFILE 1005 (n=901). In PROFILE 1005, efficacy data were analysed from a subgroup of the population in the study (n=261), whereas the safety analysis was performed on all patients. The median duration of follow-up for the subgroup was 14.2 months. The manufacturer called the subgroup the ‘mature population’; these patients had been confirmed as having ALK-positive non-small-cell lung cancer by the central laboratory and had been recruited first, so therefore the length of follow-up was longer than for the full population (the percentage of events [death] observed in this subgroup by the cut-off date, January 2012, was reported as confidential information and cannot be reported here). Endpoints for both PROFILE 1001 and PROFILE 1005 included PFS and overall survival; in addition, PROFILE 1005 assessed health-related quality of life through EQ-5D. Median PFS for the PROFILE 1005 subgroup at data cut-off (54% disease progression events observed) was 8.1 months (95% CI 6.8 to 9.7). Median overall survivalfor the subgroup was presented as confidential information and therefore cannot be reported in this document.

3.9 The manufacturer reported adverse reactions from PROFILE 1001, PROFILE 1005 and PROFILE 1007. Pooled analyses from the PROFILE studies were provided separately for hepatotoxicity and testosterone levels. In PROFILE 1001, 98% of patients (n=117) experienced an adverse reaction of any kind; 47% of patients (n=56) reported an adverse reaction with a severity of grade 3 or greater and 7% (n=8) discontinued as a result of an adverse reaction. In PROFILE 1005, the most common adverse reactions (occurring in 10% or more patients) were: gastrointestinal disorders (for example, nausea 47%,vomiting 39% and diarrhoea 41%) and visual impairment 52%. Treatment-related grade 3 or 4 adverse reactions were reported in 25.6% of patients, most frequently neutropenia (n=50 [5.5%]), increased alanine aminotransferase (n=36 [4.0%]) and fatigue (n=18 [2.0%]). Of the 198 deaths in the study, 4 were considered treatment-related by the investigators. Adverse reaction data from PROFILE 1007 were presented as confidential and cannot therefore be reported in this document.

3.10 Quality of life was measured through the European Organization for the research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) and EQ-5D questionnaire. The overall difference in change from baseline scores in global quality of life and functioning domains was found to be significantly different between the 2 treatment groups, with a statistically significantly greater improvement observed for crizotinib compared with chemotherapy (p<0.0001). In the study, EQ-5D index scores were higher in the crizotinib group than in the chemotherapy group. The mean EQ-5D index scores were presented as confidential information and cannot be reported here.

3.11 No direct comparative evidence for crizotinib compared with best supportive care was identified by the manufacturer in a systematic review, therefore a mixed treatment comparison was conducted to attempt to provide overall survivaland PFS hazard ratios for this comparison. To make a network of studies, the manufacturer stated that it was necessary to include studies that included patients with ALK-negative non-small-cell lung cancer. Four studies were identified for inclusion in the mixed treatment comparison: PROFILE 1007 (comparing crizotinib with docetaxel or pemetrexed), JMEI and GFPC 05-06 (comparing docetaxel with pemetrexed), and TAX-317 (comparing docetaxel with best supportive care). The manufacturer assessed the extent to which the differences in age, disease stage, performance status, line of treatment, crossover and non-use of intention-to-treat analysis in GFPC 05-06 would be a potential source of heterogeneity. A fixed-effects model was chosen because of the low number of studies in the network. Two analyseswere carried out that differed in how crizotinib was connected to the network; that is, using either the docetaxel (analysis 1) or pemetrexed (analysis 2) subgroup of the chemotherapy arm from PROFILE 1007 as the link with the other studies.

3.12 Data on PFS were not available for all studies in the mixed treatment comparison and therefore, overall survival was the only outcome considered. Crossover-adjusted hazard ratios were not available for crizotinib compared with the subgroups (docetaxel and pemetrexed), therefore it was assumed that the overall survival hazard ratios of crizotinib compared with docetaxel, and crizotinib compared with pemetrexed were equivalent. The overall survival hazard ratio for crizotinib compared with chemotherapy from PROFILE 1007, adjusted for crossover using the manufacturer’s preferred IPTCW approach, was used as a proxy for both comparisons.The results of the mixed treatment comparison were presented as confidential information and cannot be reported here. The manufacturer concluded that the results of analyses 1 and 2 were consistent with each other, and that crizotinib has a statistically significant effect on overall survival compared with best supportive care.

3.13 The manufacturer developed a 3-health-state model, which it referred to as a semi-Markov area-under-the-curve analysis. The 3 health states in the model wereprogression free, progressed disease and death. All patients entered the model after failure of 1 chemotherapy regimen but were considered progression freebecause they had not reached the criterion for further progression. After each 30-day cycle, patients either remainedprogression free or moved through the model to progressed disease or death. Patients in the progressed diseasehealth state received best supportive care in line with current clinical practice. Life years were generated by the model according to the proportion of patients in the progression free and progressed disease states at each cycle and summing these over the lifetime horizon (15 years in the base case). All patients were assumed to have ALK-positive non-small-cell lung cancer because testing was assumed to be performed before, or during, first-line treatment. The model provided pairwise comparisons between crizotinib and docetaxel, best supportive care or pooled chemotherapy (docetaxel and pemetrexed).

3.14 The manufacturer’s model used estimates of treatment effectiveness (PFS and overall survival) from different sources. PFS for crizotinib and docetaxel were extrapolated by fitting a Weibull distribution (crizotinib) or a log-normal distribution (docetaxel) to the Kaplan-Meier plots in PROFILE 1007; PFS for best supportive care was assumed to be equal to docetaxel in PROFILE 1007; overall survival for crizotinib was extrapolated by fitting a Weibull distribution to the Kaplan-Meier plot for the subgroup (called the ‘mature population’) from PROFILE 1005. Overall survivalfor docetaxel was estimated by applying the crossover-adjusted hazard ratio between crizotinib and pooled chemotherapy from PROFILE 1007 (see section 3.7) to the crizotinib overall survival curve (from PROFILE 1005). For best supportive care, overall survival was estimated by applying a generated hazard ratio between crizotinib and best supportive care from the mixed treatment comparison (see section 3.11) to the crizotinib overall survival curve (from PROFILE 1005). Using these data sources, the proportion of patients in the progressed disease health state was calculated as the difference between overall survival and PFS.

3.15 Resource use in the model included costs relating to drug acquisition (according to BNF 63 prices), administration (docetaxel only) and monitoring, treatingadverse reactions, ALK testing, and costs of best supportive care, routine medical management and terminal care. The average treatment duration for crizotinib came from PROFILE 1007. In the base-case analysis, treatment was assumed to stop on progression to the progressed disease health state (assessed radiographically every 6 weeks), which occurred after a median duration of 33 weeks. A scenario analysis was performed assuming treatment to the point of discontinuation observed in PROFILE 1007 (the number of weeks was presented as confidential information and cannot be reported here). A half-cycle correction was applied to the crizotinib acquisition costs, which effectively assumed no wastage was incurred by patients who moved to the progressed disease health state at the midpoint of any monthly cycle. Treatment with docetaxel was assumed to continue until disease progression. The administration cost of docetaxel was 102.11 per cycle; no administration costs were included in the total cost for crizotinib or best supportive care. The model included the cost of CT scans by health state and assumed that 30% of patients received 0.75 scans per month in the progression-free health state and 5% of patients received 0.75 scans per month in the progressed disease health state. The only adverse event considered in the model was neutropenia, rates of which were taken from PROFILE 1007. The costs of treating neutropenia were attributed to treatment with docetaxel only, because neutropenia from treatment with crizotinib was assumed not to need active treatment as crizotinib would not continue to be given. The manufacturer’s base-case testing strategy was to test the entire non-small-cell lung cancer population using immunohistochemistry (IHC) and confirm equivocal results (IHC 1+ and 2+) using a fluorescence in-situ hybridisation (FISH) test. The manufacturer calculated the cost of each testing strategy as the expected cost per patient to identify 1 patient with ALK-positive non-small-cell lung cancer from a cohort of patients with all types of non-small-cell lung cancer. The manufacturer assumed an ALK-positive non-small-cell lung cancer prevalence of 5%, resulting in a multiplication factor of 20 for testing the entire non-small-cell lung cancer cohort. The unit cost of IHC was 25 and the unit cost of a FISH test was reported as confidential information and therefore cannot be reported here. The total cost of ALK testing in the model was 630.06 for each patient receiving treatment with crizotinib.

3.16 Utility values for the progression-free and progressed disease health states were derived from EQ-5D data collected in PROFILE 1007. For the progression-free health state, utilities were calculated (for crizotinib and docetaxel separately) as the average EQ-5D index value at each time point in PROFILE 1007 weighted by the number of patients still on treatment at that time. The health state utility values were reported as confidential information and cannot be reported here. Because utility values were not available for best supportive care from the trial, these were assumed to be equal to the utility values for baseline chemotherapyfor progression-free disease and the end of treatment utility value for chemotherapy for progressed disease. Disutility for adverse reactions associated with treatment was not applied in the model.

3.17 In the manufacturer’s base-case analysis, the modelled PFS for crizotinib was 9.6 months, compared with 3.9 months for both docetaxel and best supportive care (incremental benefit for crizotinib of 5.7 months). The modelled time in the progressed disease health state was 23.4 months for crizotinib and 16.8 months for both docetaxel and best supportive care (incremental benefit for crizotinib of 6.6 months). The modelled 12-month survival probability for crizotinib was 67.7%, compared with 54.4% for docetaxel and 34.6% for best supportive care. The base-case analysis resulted in a deterministic incremental cost-effectiveness ratio (ICER) of 41,544 per quality-adjusted life year (QALY) gained for crizotinib compared with docetaxel (incremental costs of 40,227 and incremental QALYs of 0.968) and 35,455 per QALY gained for crizotinib compared with best supportive care (incremental costs of 48,128 and incremental QALYs of 1.357).

3.18 The manufacturer undertook a series of one-way deterministic sensitivity analyses to test the robustness of the results by varying several parameters used in the economic evaluation. Hazard ratios for overall survival (for crizotinib compared with docetaxel, and crizotinib compared with best supportive care) were varied between the upper and lower limits of the 95% confidence interval. Other parameters, including the utility values for progression-free and progressed disease for all treatments, administration costs of chemotherapy and the costs of diagnostic testing were varied between 20% below and 20% above the base-case value. From the results, the manufacturer highlighted that the ICER for crizotinib was very sensitive to the hazard ratios for overall survival, which affected the absolute QALYs for both the crizotinib and comparator arm. This increased the ICER from the base case of 41,544 per QALY gained to 239,699 per QALY gained in the comparison with docetaxel, and from the base case of 35,455 per QALY gained to 82,289 per QALY gained in the comparison with best supportive care. The results were also sensitive to utility values for the progressed disease and progression-free states for crizotinib, docetaxel and best supportive care: using a lower utility value for crizotinib for progressed disease (with utility values for other treatments remaining as in the base case), the ICER increased to 47,939 per QALY gained for crizotinib compared with docetaxel and 39,184 per QALY gained for crizotinib compared with best supportive care. The manufacturer also carried out a series of scenario analyses using different extrapolation methods, different estimates for utility, treatment duration and treatment effects. The most pronounced impact on the ICER was observed with a treatment duration for crizotinib based on PROFILE 1007; with this change alone, the ICER increased to 63,785 per QALY gained for crizotinib compared with docetaxel and 51,662 per QALY gained for crizotinib compared with best supportive care. The manufacturer concludedfrom the deterministic and scenario analysesthat the key driver of the cost effectiveness of crizotinib was the overall survival hazard ratios for crizotinib compared with the comparators in the model (derived from the crossover analyses and mixed treatment comparison). In addition, the model was also sensitive, although to a much lesser extent, to variations in the utility value, the method of overall survival and PFS extrapolation, treatment duration  and using a 10-year (instead of 15-year) time horizon.

3.19 The manufacturer undertook probabilistic sensitivity analyses inwhich the probability of crizotinib being cost effective at a threshold of 50,000 per QALY gained was 64% compared with docetaxel and 81% when compared with best supportive care. At a threshold of 30,000 per QALY gained, the probability fellto below 20% and 30% for the comparisons respectively.

3.20 The manufacturer indicated that crizotinib fulfils the criteria to be assessed as an end-of-life treatment, highlighting that the patient population covered by the marketing authorisation is likely to be small (around 550 patients in the UK); current expected survival is only 6–8 months and the results of the crossover-adjusted overall survivalanalysis indicated extension to life of more than 3 months.

Evidence Review Group comments

3.21 The ERG considered that the manufacturer had identified all the available evidence on the clinical effectiveness of crizotinib. Itconsidered that the major limitation of the clinical-effectiveness evidence submitted by the manufacturer was the lack of overall survival data for crizotinib. This was because of the immaturity of the available survival data. The ERG considered the manufacturer’s use of the term ‘mature’ in relation to the PROFILE 1005 subgroup to be misleading. It considered that the data could not be considered mature for either PROFILE 1007 or PROFILE 1005 given the low overall survival event rates (28% in PROFILE 1007,percentage in PROFILE 1005 presented as confidential). The ERG also considered that the higher overall survival event rate in PROFILE 1005 may reflect the poorer performance status of the patients in PROFILE 1005, rather than the maturity of the data. In addition, both studies had a similar duration (PROFILE 1005 was slightly shorter), and the median time to follow-up was only 2 months longer for PROFILE 1005 than for PROFILE 1007 (14.2 compared with 12.2 months).

3.22 The ERG raised issues relating to the comparability of the populations within PROFILE 1007, and between PROFILE 1007 and PROFILE 1005. Within PROFILE 1007, the ERG’s main concern related to the differences between patient characteristics in the chemotherapy subgroups (docetaxel and pemetrexed). Between PROFILE 1007 and PROFILE 1005, the ERG noted some significant differences in the baseline characteristics. In particular, the ERG considered that the number and type of previous therapies received and the ECOG performance status could affect treatment response.

3.23 The ERG considered there to be a lack of a randomised comparison between crizotinib and docetaxel on which to draw robust conclusions of relative efficacy. This wasbecause of the use of pemetrexed as an additional comparator in PROFILE 1007 but also as the treatment of choice in the comparator arm of the study. The ERG considered that this would lead to selection bias. Itstated that includingdata from pemetrexed in the PROFILE 1007 comparator arm would increase the power of the efficacy outcomes, but would potentially underestimate the incidence of drug-related adverse reactions (because more toxicity is associated with docetaxel than pemetrexed).

3.24 The ERG considered the manufacturer’s different approaches to adjusting for crossover. It noted that the manufacturer had excluded 2 approaches that produced estimates favouring chemotherapy as clinically implausible. The ERG challenged this justification on the basis that there are occasions where a benefit in PFS may not translate into a benefit in overall survival. Considering the manufacturer’s preferred method of adjusting for crossover (IPTCW), the ERG pointed to the limitations of observational methods such as IPTCW, which are sensitive to the proportion of patients who cross over and sample size. The ERG stated that the sample size in PROFILE 1007 is too small for the level of crossover to reliably inform overall survival estimates and therefore, it felt the manufacturer’s preferred estimates using the IPTCW method had questionable validity. The ERG questioned the plausibility of the results of the IPTCW5 (the manufacturer’s preferred crossover-adjustment method), given that the absolute survival gain with crizotinib (12.3 months) was more than double that of the PFS gain with crizotinib (5.7 months) from the crossover analysis. The ERG also raised concerns about applying the RPSFT method to the PROFILE 1007 data, because this method only produces unbiased estimates when treatment effect is not time dependent, that is, irrespective of the number of previous treatments. However, the ERG noted that the RPSFT resulted in a gain of 5.8 months in overall survival for crizotinib compared with chemotherapy, and considered that this result was consistent with what might be expected given the gain in PFS with crizotinib. Therefore the ERG considered the RPSFT to have greater face validity and to be a more appropriate choice of method of adjusting for crossover for the base-case analysis.

3.25 The ERG raised concerns about using PFS, the primary endpoint of PROFILE 1007, as a surrogate for overall survivalin the absence of mature data. The ERG highlighted the complex relationship between PFS and overall survival in cancer generally, but also in the case of lung cancer, and the ability of PFS to accurately predict overall survivalacross populations and treatments. The ERG explained that in the manufacturer’s submission, the gains in PFS were derived from fitting parametric curves to PFS data from PROFILE 1007 and the gains in overall survival were derived from using a parametric curve of overall survival data from crizotinib in PROFILE 1005 and applying a crossover-adjusted hazard ratio for docetaxel to this data. The predicted overall survivalgains for crizotinib were higher than would be expected, given the available evidence of the relationship between PFS and overall survival seen in patients with non-small-cell lung cancer. The ERG did not consider that it was plausible for post-progression survival to be prolonged greatly by crizotinib unless crizotinib was being continued after radiographical progression and was still having an effect. The ERG questioned the validity of the overall survival advantage of crizotinib compared with chemotherapy in the manufacturer’s analysis on account of the lack of overall survival benefit demonstrated so far in PROFILE 1007, the sample size and the size of the overall survival benefit compared with that of PFS.

3.26 The ERG questioned the reliability of the results from the manufacturer’s mixed treatment comparison based on major differences between the populations in the included trials.There were differences in:

  • the proportion of patients with squamous cell histology (GFPC 05-06 and JMEI included patients with squamous cell histology)
  • the median age of the populations in the non-crizotinib studies (GFPC 05-06, JMEI and TAX-317) which was 7–12 years older than in PROFILE 1007
  • the proportion of patients with adenocarcinoma (50% in the non-crizotinib studies compared with 93% in PROFILE 1007)
  • the percentage of women (30–35% in the non-crizotinib studies compared with 55% in PROFILE 1007).

3.27 The ERG noted that overall survival was the key driver of the cost effectiveness of crizotinib, and given the concerns raised previously about the maturity of the overall survival data in PROFILE 1005 (see section 3.21), the ERG considered that these concerns would have to be weighed against the potential impact of all other assumptions and approaches applied in the manufacturer’s model. With regard to extrapolation of overall survival in the manufacturer’s base-case analysis, the ERG considered that it would be equally plausible (and possibly a more appropriate choice given its better statistical fit) to use an exponential, instead of Weibull distribution, for extrapolating overall survival.

3.28 The ERG commented that the manufacturer’s assumption in the model that treatment stops on evidence of radiographic progression may not represent what happens in clinical practice. The ERG highlighted that radiographic assessment may lead to earlier identification of progression and crossover of treatment in PROFILE 1007 than would happen in clinical practice. The ERG noted that using the actual treatment duration of crizotinib in PROFILE 1007 would increase the duration of treatment and costs. In the manufacturer’s scenario analysis, using treatment duration as in PROFILE 1007, the ICER for crizotinib compared with docetaxel increased from the base case of 41,544 per QALY gained to 63,785 per QALY gained.

3.29 The ERG raised concerns about the validity of the assumptions made about health-related quality of life in the model. The ERG pointed out that the model assumed a higher progressed disease utility value for crizotinib over docetaxel (and best supportive care); that is, patients maintain a post-progression health-related quality of life benefit at the end of treatment throughout the time spent in the progressed disease health state. The ERG questioned the validity of the assumption that utility at the end of treatment reflected the average utility experienced in the entire progressed diseasehealth state. The ERG explained that the utility of progressed disease is assumed to be independent of time spent in the state. Therefore the ERG was not convinced that patients maintain a differential health-related quality of life benefit in the progressed disease state (and certainly not for the entire duration of the progressed disease state), and commented that this assumption was not justified or discussed by the manufacturer. The ERG stated that, in the base case, post-progression utility would be a key driver of cost effectiveness because the post-progression health state accounted for approximately 57% of the incremental gain in QALYs. The ERG also noted that progression-free utilities are higher than previously reported in non-small-cell lung cancer, and if these differences are real, this could reflect a finding that patients with ALK-positive disease are generally healthier than patients with ALK-negative disease. In addition, the ERG stated that the manufacturer had not adjusted for the different baseline utilities between treatments (the estimate of baseline utility was numerically higher for crizotinib than for chemotherapy), and that no justification had been given for not doing so.

3.30 The ERG commented that the drug acquisition cost for docetaxel in the model (based on the BNF 63) does not reflect the average price paid across England and Wales through procurement contracts. Using information from the Commercial Medicines Unit (Electronic Market Information Tool [eMIT]), a usage weighted average of price paid for a generic drug such as docetaxel can be estimated and would be a more appropriate source for this cost in the model. The ERG also highlighted that the manufacturer’s submission did not explicitly describe or justify the treatment duration for docetaxel (the median number and range of cycles in PROFILE 1007 was presented as confidential information and cannot be reported here). Clinical advisers to the ERG suggested that it is unlikely that more than 6 cycles of treatment with docetaxel would be administered in clinical practice. Consequently, the ERG raised concern that the costs of docetaxel would be overestimated in the manufacturer’s analysis.

3.31 The ERG highlighted that the manufacturer had applied a half-cycle correction to the costs of crizotinib in the model. This implied that there is no wastage associated with crizotinib, that unused tablets would be reclaimed from patients who have progressed at the midpoint of any treatment cycle. The ERG considered that, by removing the half-cycle correction, it was possible to explore the influence of this assumption on the results of the model. This was undertaken in alternative scenario analyses (see section 3.34). The ERG’s preferred assumption was to remove the half-cycle correction for the costs of crizotinib from the model.

3.32 The ERG accepted that the primary target population for testing for ALK would be patients with non-small-cell lung cancer with adenocarcinoma histology, and particularly with EGFR-negative disease. However, the ERG noted that ALK fusion can be observed with non-small-cell lung cancer of a different histology. Clinical advisers to the ERG advised that it is likely that in the UK, only patients with adenocarcinoma would undergo ALK testing. However, histopathologists advising the ERG expressed a desire to see more widespread testing to thereby include all patients with ALK fusion. The ERG estimated that approximately 100 patients would be missed if screening is confined to adenocarcinoma patients, although it cautioned that there is considerable uncertainty around this estimate. With regards to the type of test, the ERG considered it reasonable to suggest the use of IHC with subsequent confirmation of equivocal results using ALK FISH testing. However, the ERG stressed the lack of clarity as to the timing of thetesting strategy (because of uncertainty about the prevalence of the ALK fusion gene in non-adenomatous non-small-cell lung cancer) and highlighted that this would impact on the cost-effectiveness estimates.

3.33 The ERG explored an alternative approach to modelling overall survival. It noted that the manufacturer had considered proportional hazards models, namely exponential, Weibull and Gompertz models. The ERG agreed with the manufacturer that the Gompertz distribution did not provide an adequate fit to the data. However, the Weibull and exponential functions appeared to offer a similar fit. The ERG commented that because there is a lack of evidence to assess external validity, it may have been more appropriate to choose the exponential distribution given the higher internal validity (a better statistical fit) than the Weibull function selected by the manufacturer. Using the exponential survival function in the base case, the ICER for crizotinib compared with docetaxel would increase from 41,554 to 44,643 per QALY gained.

3.34 The ERG carried out exploratory analyses using alternative parameters and assumptions for 5 areas of uncertainties in the manufacturer’s model that the ERG considered to be drivers of cost effectiveness. These were: the method of adjusting for crossover in the hazard ratio estimates of overall survival (comparing the ERG’s preferred method of RPSFT with IPTCW5), crizotinib treatment duration (using the same duration of treatment as in PROFILE 1007, rather than the manufacturer’s base-case assumption of treatment until RECIST progression), post-progression utility values (using the same utility values for all treatments), the acquisition cost of docetaxel (using the cost as reported in eMIT rather than the BNF 63) and removing the half-cycle correction for crizotinib acquisition costs. Using different combinations of these amendments, the ICER for crizotinib compared with docetaxel ranged between 63,770 per QALY gained (using the IPTCW5 crossover-adjustment method, assuming crizotinib treatment until disease progression, the BNF 64 price for docetaxel and the crizotinib utility value for both treatments after progression) and 204,315 per QALY gained (using the RFSPTcrossover-adjustment method, crizotinib treatment duration as in PROFILE 1007, the eMIT price for docetaxel and crizotinib post-progression utility values for both treatments). Substituting the crizotinib post-progression utility values with pooled chemotherapy post-progression utility values, the ICER was 181,095 per QALY gained. The ICER for crizotinib compared with best supportive care ranged between 46,824 per QALY gained (using the IPTCW5 crossover-adjustment method, assuming treatment until disease progression and assuming a pooled chemotherapy post-progression utility value for both treatments) and 80,535 per QALY gained (using the RPSFT crossover-adjustment method, crizotinib treatment as per PROFILE 1007 and pooled chemotherapy post-progression utility).

3.35 Full details of all the evidence are in the manufacturer’s submission and the ERG report, which are available from www.nice.org.uk/guidance/TAXXX

4 Consideration of the evidence

4.1 The Appraisal Committee reviewed the data available on the clinical and cost effectiveness of crizotinib, having considered evidence on the nature of non-small-cell lung cancer and the value placed on the benefits of crizotinib by people with the condition, those who represent them and clinical specialists. It also took into account the effective use of NHS resources.

4.2 The Committee heard from clinical specialists and patient experts that there are limited treatment options for people with non-small-cell lung cancer whose disease has progressed following chemotherapy. It heard from the patient experts and clinical specialists that non-small-cell lung cancer associated with an anaplastic lymphoma kinase (ALK) fusion gene is an uncommon subtype of non-small-cell lung cancer and noted the views of the patient experts and clinical specialists on the severity of the disease. The Committee also heard from the clinical specialists and patient experts that people with ALK-positive non-small-cell lung cancer particularly value the availability of an effectivetargeted therapy and the convenience of an oral formulation; neither of these features apply to docetaxel. It also heard from the clinical specialists that the side effects associated with crizotinib would be tolerated by most patients. The Committee concluded that crizotinib offers potential benefits to people with ALK-positive non-small-cell lung cancer.

4.3 The Committee discussed the decision problem as presented in the manufacturer’s submission. It noted that this was the same as the scope for the appraisal, with the exception that the appraisal scope listed erlotinib as a comparator, but the manufacturer had not included a comparison of  crizotinib and erlotinib in the submission. The Committee understood that erlotinib is a treatment that targets the activated epidermal growth factor receptor (EGFR) gene mutation in non-small-cell lung cancer and that it is very rare for people with non-small-cell lung cancer to have both the EGFR mutation and ALK fusion gene. It therefore accepted the manufacturer’s position that an EGFR-targeted medicine would not be expected to be standard of care in clinical practice for patients with ALK-positive disease. Therefore, the Committee agreed that erlotinib should not be considered as a comparator for crizotinib for treating previously treated ALK-positive non-small-cell lung cancer. It also noted that pemetrexed was not in the scope and not a valid comparator as a second-line treatment because patients are likely to have been treated with pemetrexed before being considered for crizotinib. The Committee was also aware that pemetrexed is not recommended by NICE as a second-line treatment. It concluded that docetaxel and best supportive care were the appropriate comparators for crizotinib.

4.4 The Committee discussed the characteristics of the population in PROFILE 1007. It noted that most of the trial population had been diagnosed with adenocarcinoma, had a good performance status, were relatively young and had never smoked. The Committee considered that these characteristics generally indicate better prognosis and therefore discussed whether the trial population represented people with ALK-positive non-small-cell lung cancer in clinical practice. It heard from the clinical specialists that the modest benefits of docetaxel in PROFILE 1007 were consistent with what would be expected in clinical practice. The Committee noted the lack of evidence available to either ascertain the survival of patients with ALK-positive disease who had not received treatment with crizotinib or to assess the separate impact on survival of the features of non-small-cell lung cancer that accompany ALK-positive disease (young age, mainly women, nearly always adenocarcinoma, and a high proportion of people who have never smoked). Although it questioned whether such patients might have a better prognosis than patients with ALK-negative disease on account of these favourable prognostic factors, the Committee accepted that the PROFILE 1007 population was likely to be similar to people considered for treatment with crizotinib in UK clinical practice.

4.5 The Committee considered treatment duration with crizotinib. It noted that a high proportion of patients in PROFILE 1007 continued to receive treatment with crizotinib after radiographically determined disease progression. It noted that the summary of product characteristics states that ‘prolongation of treatment after objective disease progression in selected patients may be considered on an individual basis, but no additional benefit has been demonstrated’. The Committee discussed whether treatment would be discontinued on radiographic disease progression in clinical practice. It heard from the clinical specialists that in practice, if a patient’s disease has previouslyresponded to treatment and the patient has not experienced adverse reactions, symptomatic progression rather than radiographic progression is likely to be the trigger for treatment change or discontinuation. The Committee were informed of  an abstract presented at the American Society of Clinical Oncology in which it was reported that 53% of patients in PROFILE 1001 and PROFILE 1005 received crizotinib post-progressed disease for at least 2 weeks (range 2–84 weeks, median 10 weeks). The Committee therefore concluded that the treatment protocol of PROFILE 1007, in which patients could continue treatment after radiographic progression, reflected the likely treatment duration for crizotinib in UK clinical practice.

4.6 The Committee discussed the evidence for the clinical efficacy of crizotinib. It noted the median gains in PFS of 4.7 and 5.1 months with crizotinib compared with chemotherapy and docetaxel respectively from PROFILE 1007, and considered that this represented a noteworthyextension to PFS in advanced non-small-cell lung cancer. It noted the objective response rate of around 65% and considered this to be a very high response rate for a second-line non-small-cell lung cancer treatment. The Committee went on to discuss the overall survival estimates from PROFILE 1007. It noted that the results did not identify a statistically significant difference in overall survival between crizotinib and chemotherapy. However,the Committee acknowledged that this was based on relatively immature data and subject to a high rate of crossover from chemotherapy to crizotinib. It heard from the manufacturer that more mature and therefore more reliable overall survival data would be available for PROFILE 1007 at [confidential information removed on 25/03/13 at the request of the company]; however, it noted that this would not be within the timeframe of this appraisal. The Committee therefore considered the results of the manufacturer’s crossover analyses in which the estimate of overall survival gain with crizotinib ranged from 5.8 months to 21.7 months. The Committee considered that the range of results from the crossover analyses suggested a high degree of uncertainty around the estimate of overall survivalgain. It heard from the clinical specialists that the estimated gain in overall survivalwith treatment might be expected to be 8 or 9 months. The Committee noted that this was approximately midway between the results of the RPSFT method and the manufacturer’s chosen method (IPTCW5). It therefore accepted that treatment with crizotinib would result in an overall survival gain compared with docetaxel but the exact magnitude of the gain was uncertain because of the immaturity of the PROFILE 1007 data and the impact of crossover in the study. Overall, the Committee concluded that, on the basis of the evidence for PFS and response rate, crizotinib is a clinically efficacious treatment for ALK-positive advanced  non-small-cell lung cancer.

4.7 The Committee noted the number of adverse events associated with crizotinib treatment from the PROFILE studies (see section 3.9). However, it was advised by the patient experts and clinical specialists that crizotinib would be tolerated by most people with non-small-cell lung cancer. The Committee concluded that crizotinib is associated with some adverse reactions but these would be tolerable for most patients and generally easily managed.

4.8 The Committee discussed the results of the manufacturer’s mixed treatment comparison in which crizotinib was compared with best supportive care. It noted the Evidence Review Group’s (ERG’s)assertion that there were substantial underlying differences in the included studies in the populations of patients with non-small-cell lung cancer. It further noted the ERG’s concern that the populations in the mixed treatment comparison were well enough for chemotherapy, and therefore the prognostic factors of these populations would not be representative of those receiving best supportive care. Therefore the Committee concluded that the results of the mixed treatment comparison were subject to very significant heterogeneity in the included studies and the results should be viewed with considerable caution.

4.9 The Committee discussed the manufacturer’s preferred approach to crossover (IPTCW5) in more detail, noting that this had been used to obtain the overall survival hazard ratio for docetaxel. It was aware that the ERG considered the RPSFT method to be more methodologically robust than the IPTCW method because RPSFT maintains randomisation. In addition, the IPTCW approach is sensitive to both the proportion of patients who cross over (increased bias with increased proportion) and the sample size (increased bias with smaller sample size). Given the high percentage of people that crossed over from chemotherapy to receive crizotinib in PROFILE 1007, the Committee agreed that the analysis would be affected by these concerns. The Committee was also aware that the clinical specialists had indicated that there was no evidence to suggest that the benefit of crizotinib was very different when used second or third line. This therefore gave more credence to the RPSFT method of adjusting for crossover. The Committee further heard from the ERG that a recent technical support document by the NICE Decision Support Unit had found that in most cases examined, the RPSFT method had been the most methodologically appropriate approach. The Committee questioned the manufacturer’s justification for preferring one method, noting that of the different statistical methods, IPTCW5 gave the most favourable overall survival benefit for crizotinib. It heard from the manufacturer that the decision was based on their view that the resulting hazard ratio from the IPTCW5 method when applied to the extrapolated overall survival for crizotinib most closely reflected the overall survival from other trials of docetaxel and pemetrexed. The Committee noted that the chosen method resulted in a modelled PFS gain of 5.7 months, and an overall survival gain of 12.3 months for crizotinib. It considered that an overall survival gain more than twice the size of a gain in PFS was unlikely. Furthermore, it noted that the other trials of second-line treatment with pemetrexed or docetaxel were in potentially very different populations of patients. It therefore did not accept the manufacturer’s assertion of face validity. The Committee also challenged the use of the ‘last observation carried forward’ for 56 days method for imputing missing data, noting that there are alternative methods to imputing data, but that the manufacturer had not justified the selection of this particular method for dealing with missing data. The Committee went on to question the application of the manufacturer’s method of crossover, questioning why the type of chemotherapy had not been included as a covariant, given that pemetrexed had been given as the first choice treatment in the chemotherapy group. It heard from the manufacturer that this had not been considered. The Committee agreed that the manufacturer’s application of the chosen method for adjusting for crossover (IPCTW5) was highly uncertain. It considered the outcome from the RPSFT method, in which the overall survival benefit for crizotinib was 5.8 months. In view of the evidence from the clinical specialists relating to the expected gain in survival with crizotinib (see section 4.6), the Committee concluded that the RPSFT method might have resulted in an underestimate of overall survival. However, it agreed that the manufacturer’s estimate of 12.3 months was implausibly high because it was greater than double the PFS gain observed. In addition the Committee considered the uncertainty resulting from the application of the crossover methods was compounded by the immaturity of the data on which all the analyses were based. The Committee concluded that the exact gain in overall survival from treatment with crizotinib was uncertain, but on balance, it would be nearer the estimate produced by the RPSFT method of adjusting for crossover than the IPCTW5 method.

4.10 The Committee discussed the utility estimates in the model. It welcomed the collection of EQ-5D data in PROFILE 1007. The Committee noted that the baseline utility estimates were different between the groups at entry into the study, and specifically that the mean baseline utility value for crizotinib was higher than for chemotherapy. The manufacturer confirmed that this had not been adjusted for in the model. The Committee also noted the difference in utility values between crizotinib and chemotherapy for the progressed disease health state and observed that these post-progression utilities had been measured at the outset of the progressed disease state and continued at that value until death. It first discussed whether a benefit of treatment with crizotinib might be expected to continue beyond treatment being discontinued. The Committee heard from the clinical specialists that patients with progressed disease would continue to receive some additional health-related quality of life benefit for some time after treatment was withdrawn compared with those on chemotherapy, but that this would deteriorate over time. It accepted that some utility benefit might be expected from crizotinib discontinued at disease progression, though there are no data to suggest how great a benefit this might be or for how long it would persist. The Committee was also aware that there might be a benefit to utility of actively continuing crizotinib but there was no data to show whether such continued treatment benefits patients or for how long. It heard from the ERG that the manufacturer could have carried out a sensitivity analysis on the differential post-progression utility values to explore the effect of this uncertainty on the cost effectiveness of crizotinib. The Committee concluded that an initial difference in post-progression utility is plausible, but that this would reduce rather than persist throughout the progressed disease state.

4.11 The Committee discussed the cost estimates in the manufacturer’s economic model. It noted that scans were given every 6 weeks in PROFILE 1007. The Committee heard from the clinical specialists that on average, patients would initially receive a CT scan every 2 months and this would probably be reduced to every 3 months at a later stage if the patient was clearly benefitting. The Committee considered that the costs of CT scans in the model in which the Committee calculated the assumption to represent approximately 1 scan every 4 cycles in the progression free health state, had therefore been underestimated for crizotinib. The Committee noted the costs of docetaxel in the model were based on its use in the post hoc subgroup in PROFILE1007 (presented as confidential information in the manufacturer’s submission and not reported here). The Committee thought it very unlikely that in England and Wales, patientswould receive more than 6 cycles of docetaxel. It therefore considered that this cost for docetaxel had been overestimated in the manufacturer’s model. The Committee considered the administration costs, noting that the model assumed no cost to the NHS associated with administration of crizotinib. It agreed that there would be some administrative costs to the NHS associated with treatment with crizotinib and that the SB11Z healthcare resource group code for oral chemotherapy of 126 should have been included for each crizotinib treatment cycle in the progression-free state. The Committee also considered drug wastage, noting that the manufacturer’s application of the half-cycle correction assumed that no wastage would be incurred. However, it heard from the clinical specialists that once a pack of tablets has been dispensed, any unused tablets cannot be returned and so it is inevitable that there may be some wastage. The Committee therefore considered it appropriate to remove the half-cycle correction from the model. Finally, the Committee considered the acquisition price for docetaxel, noting the substantial discrepancy between the published price in the ‘British National Formulary’ (BNF) and the range of prices paid by the NHS across the country as reported in the electronic Market Information Tool (eMIT) from the Commercial Medicines Unit of the NHS. Overall, the Committee agreed that the costs in the model were likely to be underestimated in favour of crizotinib because of the more regular use of CT scans in clinical practice, the lack of administration cost, the use of a half-cycle correction and overestimation of the costs ofdocetaxel because of the number of cycles likely to be administered and its acquisition cost. The Committee concluded that the impact of these factors would increase the ICER in the manufacturer’s model.

4.12 The Committee considered the most plausible cost-effectiveness estimates of crizotinib compared with docetaxel and best supportive care. Based on its discussions about the crossover-adjusted survival estimates for crizotinib compared with chemotherapy (see section 4.9), the Committee considered any estimates of the ICER based on the IPTCW5 method would be an underestimate. It further considered that the manufacturer’s base-case ICERs were based on an underestimate of the true costs associated with crizotinib and an overestimate of the true costs of docetaxel (see section 4.11). It further considered that the manufacturer’s ICER would unduly favour crizotinib because of the assumption of a continued benefit in utility in progressed disease (see section 4.10). For the comparison with best supportive care, the Committee considered that the ICER would be affected by most of these concerns (except the underestimate of the costs of docetaxel) and in addition, by the uncertainty in the mixed treatment comparison results (see section 4.8). The Committee concluded that it could not base its decision on the manufacturer’s base-case estimates of the ICER.

4.13 The Committee further considered the most plausible cost-effectiveness estimates of crizotinib compared with docetaxel and best supportive care. It noted the ERG’s exploratory analysis that used the RPSFT method for crossover and addressed some of the Committee’s concerns, namely treatment duration as in PROFILE 1007, removal of the half-cycle correction for crizotinib and use of eMIT-reported acquisition costs for docetaxel. It understood that, assuming a pooled chemotherapy post-progression utility for both treatment strategies, this analysis resulted in an ICER for crizotinib compared with docetaxel of 181,100 per QALY gained. However, it recognised that this ICER assumed no difference in utility between crizotinib and the comparator treatments during progressed disease, and on the basis of its discussions around this point (see section 4.10), it considered that most plausible ICER would therefore be somewhat lower than this. The Committee noted the manufacturer’s scenario analysis in which the base-case assumption of treatment duration was amended to reflect that of PROFILE 1007 provided an ICER estimate of 63,800 per QALY gained for crizotinib compared with docetaxel, and 51,700 per QALY gained for crizotinib compared with best supportive care. The Committee concluded that the most plausible ICERs would lie somewhere between the range of the 2 estimates, that is, between 63,800 and 181,100 per QALY gained for crizotinib compared with docetaxel, and between 51,700 and 80,500 per QALY gained for crizotinib compared with best supportive care.

4.14 The Committee considered whether crizotinib offers benefits because of its innovative nature, as the first targeted drug for ALK-positive non-small-cell lung cancer. It heard from the manufacturer that crizotinib is innovative because the ability to target patients who are most likely to benefit can be seen as a step change in the management of non-small-cell lung cancer. It further heard from the clinical specialists and patient experts that crizotinib delivers high response rates and a substantial benefit in at least progression-free survival in lung cancer and is also well tolerated, in particular when compared with current standard cytotoxic therapy for non-small-cell lung cancer. The Committee agreed with these observations but considered that the potential extension to life and the convenience of an oral treatment compared with intravenous second-line therapy would already be captured in the QALY calculation, and concluded that no additional value judgements neededto be made for innovation.

4.15 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 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 that the assumptions used in the reference case of the economic modelling are plausible, objective and robust.

4.16 The Committee considered the life expectancy of patients with advanced non-small-cell lung cancer associated with an ALK fusion gene. It noted the results from the manufacturer’s statistical crossover analyses, which gave a range of estimates between 20 and 27 months for the chemotherapy group. On the basis of its discussions around the crossover methods explored by the manufacturer (see section 4.9), it considered that there was some uncertainty around these estimates. It further acknowledged that there is a lack of survival data for patients with ALK-positive non-small-cell lung cancer who have not received treatment with crizotinib. However, on balance, the Committee considered that the life expectancy of people with ALK-positive non-small-cell lung cancer after first-line chemotherapy would be less than 24 months. It then discussed the criterion relating to extension to life. The Committee noted that the median PFS results from PROFILE 1007 indicated an extension to life of 4.7 months for crizotinib compared with chemotherapy, and that this was not affected by crossover. It agreed that crizotinib would extend life by an additional 3 months compared with chemotherapy. The Committee then considered the size of the population, noting the manufacturer’s estimate of around 500 patients. It accepted crizotinib is licensed for a small population. The Committee accepted that the supplementary advice from NICE for life-extending treatments could be considered for crizotinib given the above three criteria. The Committee then considered its duty to be satisfied that the estimates to the extension of life and the assumptions used in the economic modelling were plausible, objective and robust, in order for the supplementary advice to be used in the context of the decision to be made. It did not agree that the assumptions used in the economic model were plausible, objective or robust. Given this, the Committee was unable to consider the magnitude of the additional weight that would need to be assigned to the QALY benefits for the cost effectiveness of crizotinib to fall within the current threshold range.

4.17 The Committee considered its recommendation for guidance to the NHS. On the basis of the most plausible ICERs (see sections 4.12 and 4.13), the Committee noted that even if the supplementary advice to the Committee for life-extending treatments had applied, crizotinib could not be considered a cost-effective use of NHS resources. It therefore concluded that treatment with crizotinib for previously treated ALK-positive advanced non-small-cell lung cancer should not be recommended for use within the NHS.

4.18 The Committee considered whether its recommendations were associated with any potential issues related to equality. The Committee noted the potential equalities issue raised during scoping that testing could be restricted to those patients with a diagnosis of adenocarcinoma. The Committee heard from the clinical specialists that there is currently no established ALK testing strategy in UK clinical practice. The Committee discussed whether this potential equalities issue impacted on NICE’s duties under the equalities legislation and agreed that the preliminary recommendations did not differentiate between any groups of people. It concluded that its recommendations do not have a particular impact on any of the groups whose interests are protected by the legislation and that there was no need to alter or add to its recommendations.

Summary of Appraisal Committee’s key conclusions

TAXXX Appraisal title: Crizotinib for treating previously treated non-small-cell-lung cancer associated with an anaplastic lymphoma kinase fusion gene Section
Key conclusion

On the basis of the most plausible ICERs, the Committee noted that even if the supplementary advice to the Committee for life-extending treatments had applied, crizotinib could not be considered a cost-effective use of NHS resources. It therefore concluded that treatment with crizotinib for previously treated ALK-positive advanced non-small-cell lung cancer should not be recommended for use within the NHS.

The Committee accepted that treatment with crizotinib would result in an OS survival gain compared with docetaxel but the exact magnitude of the survival gain was uncertain because of the immaturity of the 1007 trial data and the impact of crossover in the trial.

The Committee concluded that, on the basis of the evidence for PFS and response rate, crizotinib is a clinically efficacious treatment for NSCLC associated with the ALK fusion gene.

The Committee noted the manufacturer’s sensitivity analyses using treatment duration as per the PROFILE 1007 trial which provided an ICER estimate of 63,800 per QALY gained for crizotinib compared with docetaxel and 51,700 per QALY gained for crizotinib compared with best supportive care. The Committee concluded that these estimates would represent the lowest possible estimate of the ICERs.

4.17

4.6

4.6

4.13

Current practice
Clinical need of patients, including the availability of alternative treatments

The Committee heard from clinical specialists and patient experts that there are limited treatment options for people with non-small-cell lung cancer whose disease has failed chemotherapy.

The Committee also heard from the clinical specialists and patient experts that people with ALK-positive non-small-cell lung cancer particularly value the availability of an effective targeted therapy and the convenience of an oral formulation; neither of these features apply to docetaxel.

4.2
The technology

Proposed benefits of the technology

How innovative is the technology in its potential to make a significant and substantial impact on health-related benefits?

The Committee considered whether crizotinib offers benefits because of its innovative nature, as the first targeted drug for ALK-positive non-small-cell lung cancer. It heard from the manufacturer that crizotinib is innovative because the ability to target patients who are most likely to benefit can be seen as a step change in the management of non-small-cell lung cancer. It further heard from the clinical experts and patient specialists that crizotinib delivers a substantial benefit in at least progression-free survival in lung cancer and is well tolerated, in particular when compared with current standard cytotoxic therapy for non-small-cell lung cancer. 4.14
What is the position of the treatment in the pathway of care for the condition? The Committee concluded that docetaxel and best supportive care were the appropriate comparators for crizotinib. . 4.3
Adverse reactions The Committee noted the number of adverse events associated with crizotinib treatment from the PROFILE studies. The Committee concluded that crizotinib is associated with some adverse reactions but these would be tolerable for most patients and generally easily managed. 4.7
Evidence for clinical effectiveness
Availability, nature and quality of evidence The main evidence came from 1 multi-centre, randomised phase III efficacy and safety study in patients with previously treated ALK-positive non-small-cell lung cancer (PROFILE 1007). 3.2
Relevance to general clinical practice in the NHS

The Committee accepted that the PROFILE 1007 population was likely to be similar to people considered for treatment with crizotinib in UK clinical practice.

The Committee concluded that the treatment protocol of PROFILE 1007, in which patients could continue treatment after radiographic progression, reflected the likely treatment duration for crizotinib in UK clinical practice.

4.4

4.5

Uncertainties generated by the evidence The Committee acknowledged that the overall survival data from the crizotinib studies was relatively immature and, in the case of PROFILE 1007, subject to a high rate of crossover from chemotherapy to crizotinib. The Committee heard from the manufacturer that more mature and therefore more reliable overall survival data would be available for PROFILE 1007 at [confidential information removed on 25/03/13 at the request of the company]; however it noted that this would not be within the timeframe of this appraisal. 4.6
Are there any clinically relevant subgroups for which there is evidence of differential effectiveness? Subgroups for patients receiving treatment with crizotinib were not in the scope, nor identified during the appraisal.  
Estimate of the size of the clinical effectiveness including strength of supporting evidence

The Committee noted the median gain in PFS of 5.1 months with crizotinib compared with docetaxel from the PROFILE 1007 study, and considered that this represented a noteworthy extension to PFS in advanced non-small-cell lung cancer.

The Committee accepted that treatment with crizotinib would result in an OS survival gain compared with docetaxel but the exact magnitude of the survival gain was uncertain because of the immaturity of the 1007 trial data and the impact of crossover in the trial.

Overall, the Committee concluded that, on the basis of the evidence for PFS and response rate, crizotinib is a clinically efficacious treatment for ALK-positive non-small-cell lung cancer.

4.6

4.6

4.6

Evidence for cost effectiveness
Availability and nature of evidence The manufacturer developed a 3-health state model, which it referred to as a semi-Markov area-under-the-curve analysis. The model used estimates of treatment effectiveness from PROFILE 1005, PROFILE 1007 and a mixed treatment comparison. 3.13, 3.14
Uncertainties around and plausibility of assumptions and inputs in the economic model

The Committee agreed that the manufacturer’s application of the chosen method for adjusting for crossover (IPCTW5) was highly uncertain. It considered the outcome from the RPSFT method, in which the overall survival benefit for crizotinib was 5.8 months. In view of the evidence from the clinical specialists relating to the expected gain in survival with crizotinib, the Committee concluded that the RPSFT method might have resulted in an underestimate of overall survival. However, it agreed that the manufacturer’s estimate of 12.3 months was implausibly high because it was greater than double the PFS gain observed. In addition the Committee considered the uncertainty about the results from any cross-over adjustment was compounded by the immaturity of the data on which all the analyses were based. The Committee concluded that the exact gain in overall survival from treatment with crizotinib was uncertain, but on balance, it would be nearer the estimate produced by the RPSFT method of adjusting for crossover than the IPCTW5 method.

The Committee concluded that the results of the mixed treatment comparison (for crizotinib compared with best supportive care) were subject to very significant heterogeneity in the included studies and the results should be viewed with considerable caution.

4.9

4.8

Incorporation of health-related quality-of-life benefits and utility values

Have any potential significant and substantial health-related benefits been identified that were not included in the economic model, and how have they been considered?

The Committee discussed the utility estimates in the model. It noted that the baseline utility estimates were different between the groups at entry into the study, and specifically that the mean baseline utility value for crizotinib was higher than for chemotherapy. The manufacturer confirmed that this had not been adjusted for in the model.

The Committee also noted the difference in utility values for the progressed disease health state between crizotinib and chemotherapy and observed that these post-progression utilities had been measured at the outset of the progressed disease state and continued at that value until death. The Committee accepted that some utility benefit might be expected from crizotinib discontinued at disease progression, though there are no data to suggest how great a benefit this might be or for how long it would persist. The Committee concluded that an initial difference in post-progression utility is plausible, but that this would reduce rather than persist throughout the progressed disease state.

4.10
Are there specific groups of people for whom the technology is particularly cost effective? Not applicable  
What are the key drivers of cost effectiveness?

The Committee considered the most plausible cost-effectiveness estimates of crizotinib compared with docetaxel and best supportive care. On the basis of its deliberations regarding the crossover adjusted survival estimates for crizotinib compared with chemotherapy, the Committee considered any estimates of the ICER based on the IPTCW5 method would be an underestimate. It further considered that the manufacturer’s base-case ICERs were based on an underestimate of the true costs associated with crizotinib and an over-representation of the true costs of docetaxel. It further considered that the manufacturer’s ICER would unduly favour crizotinib because of the assumption of a continued benefit in utility during progressed disease.

For the comparison with best supportive care, the Committee considered that the ICER would be affected by most of these concerns (except the underestimate of the costs of docetaxel) and in addition, by the uncertainty in the mixed treatment comparison results.

4.13
Most likely cost-effectiveness estimate (given as an ICER) The Committee concluded that the most plausible ICERS would lie somewhere between 64,000 and 181,100 per QALY gained for crizotinib compared with docetaxel, and between 52,000 and 80,500 per QALY gained for crizotinib compared with best supportive care. 4.13
Additional factors taken into account
Patient access schemes (PPRS) Not applicable  
End-of-life considerations  The Committee accepted that the supplementary advice from NICE for life-extending treatments could be considered for crizotinib. The Committee then considered its duty to be satisfied that the estimates to the extension of life and the assumptions used in the reference case economic modelling were plausible, objective and robust, in order for the supplementary advice to be used in the context of the decision to be made. It did not agree that the economic modelling was robust, nor that the assumptions used in the economic model were plausible, objective or robust. Given this, the Committee was unable to consider the magnitude of the additional weight that would need to be assigned to the QALY benefits for the cost effectiveness of crizotinib to fall within the current threshold range. 4.16
Equalities considerations and social value judgements It concluded that its recommendations do not have a particular impact on any of the groups whose interests are protected by the legislation and that there was no need to alter or add to its recommendations. 4.18
       

5 Implementation

5.1 The Secretary of State and the Welsh Assembly Minister for Health and Social Services have issued directions to the NHS in England and Wales on implementing NICE technology appraisal guidance. When a NICE technology appraisal recommends use of a drug or treatment, or other technology, the NHS must usually provide funding and resources for it within 3 months of the guidance being published. If the Department of Health issues a variation to the 3-month funding direction, details will be available on the NICE website. When there is no NICE technology appraisal guidance on a drug, treatment or other technology, decisions on funding should be made locally.

5.2 The technology in this appraisal may not be the only treatment for non-small-cell lung cancer. If a NICE technology appraisal recommends use of a technology, it is as an option for the treatment of a disease or condition. This means that the technology should be available for a patient who meets the clinical criteria set out in the guidance, subject to the clinical judgement of the treating clinician. The NHS must provide funding and resources (in line with section 5.1) when the clinician concludes and the patient agrees that the recommended technology is the most appropriate to use, based on a discussion of all available treatments.

5.3 NICE has developed tools to help organisations put this guidance into practice (listed below). These are available on our website (www.nice.org.uk/guidance/TAXXX). [NICE to amend list as needed at time of publication]

  • Slides highlighting key messages for local discussion.
  • Costing template and report to estimate the national and local savings and costs associated with implementation.
  • Implementation advice on how to put the guidance into practice and national initiatives that support this locally.
  • A costing statement explaining the resource impact of this guidance.
  • Audit support for monitoring local practice.

6 Related NICE guidance

Published

7 Proposed date for review of guidance

7.1 NICE proposes that the guidance on this technology is considered for review by the Guidance Executive in May 2016. NICE welcomes comment on this proposed date. The Guidance Executive will decide whether the technology should be reviewed based on information gathered by NICE, and in consultation with consultees and commentators.

Peter Clark
Chair, Appraisal Committee
March, 2013

Appendix A: Appraisal Committee members and NICE project team

A Appraisal Committee members

The Appraisal Committees are standing advisory committees of NICE. Members are appointed for a 3-year term. A list of the Committee members who took part in the discussions for this appraisal appears below. There are four Appraisal Committees, each with a chair and vice chair. Each Appraisal Committee meets once a month, except in December when there are no meetings. Each Committee considers its own list of technologies, and ongoing topics are not moved between Committees.

Committee members are asked to declare any interests in the technology to be appraised. If it is considered there is a conflict of interest, the member is excluded from participating further in that appraisal.

The minutes of each Appraisal Committee meeting, which include the names of the members who attended and their declarations of interests, are posted on the NICE website.

Professor Peter Clark (Chair)
Consultant Medical Oncologist, Clatterbridge Centre for Oncology

Professor Jonathan Michaels (Vice Chair)
Professor of Clinical Decision Science, University of Sheffield

Professor Kathryn Abel
Director of Centre for Women’s Mental Health, University of Manchester

Dr Aomesh Bhatt
Regulatory and Medical Affairs Director Europe and North America, Reckitt Benckiser

Dr Andrew Black
General Practitioner, Mortimer Medical Practice, Herefordshire

Dr Ian Campbell
Honorary Consultant Physician, Llandough Hospital, Cardiff

Professor Usha Chakravarthy
Professor of Ophthalmology and Vision Sciences, The Queen’s University of Belfast

Tracey Cole
Lay Member

Dr Ian Davidson
Lecturer in Rehabilitation, University of Manchester

John Dervan
Lay Member

Dr Martin Duerden
Assistant Medical Director, Betsi Cadwaladr University Health Board, North Wales

Gillian Ells
Prescribing Advisor – Commissioning, NHS Hastings and Rother and NHS East Sussex Downs and Weald

Professor Paula Ghaneh
Professor and Honorary Consultant Surgeon, University of Liverpool

Dr Susan Griffin
Research Fellow, Centre for Health Economics, University of York

Professor Carol Haigh
Professor in Nursing, Manchester Metropolitan University

Professor Peter Jones
Emeritus Professor of Statistics, Keele University

Professor Femi Oyebode
Professor of Psychiatry and Consultant Psychiatrist, The National Centre for Mental Health

Dr Brian Shine
Consultant Chemical Pathologist, John Radcliffe Hospital, Oxford

Dr Surinder Sethi
Consultant in Public Health Medicine, North West Specialised Services Commissioning Team, Warrington

Dr Murray Smith
Associate Professor in Social Research in Medicines and Health, University of Nottingham

Charles Waddicor
Chief Executive, NHS Berkshire West

B NICE project team

Each technology appraisal is assigned to a team consisting of one or more health technology analysts (who act as technical leads for the appraisal), a technical adviser and a project manager.

Helen Tucker
Technical Lead

Joanne Holden
Technical Adviser

Kate Moore
Project Manager

Appendix B: Sources of evidence considered by the Committee

A The Evidence Review Group (ERG) report for this appraisal was prepared by the Centre for Reviews and Dissemination/Centre for Health Economics:

  • Duarte A, Burch J, Smith A, et al. Crizotinib for the treatment of previously treated non-small-cell lung cancer associated with an anaplastic lymphoma kinase (ALK) fusion gene. A Single Technology Appraisal, February 2013

B The following organisations accepted the invitation to participate in this appraisal as consultees and commentators. They were invited to comment on the draft scope, the ERG report and the appraisal consultation document (ACD). Organisations listed in I were also invited to make written submissions. Organisations listed in II and III had the opportunity to give their expert views. Organisations listed in I, II and III also have the opportunity to appeal against the final appraisal determination.

I Manufacturer/sponsor:

  • Pfizer

II Professional/specialist and patient/carer groups:

  • Roy Castle Lung Cancer Foundation
  • British Thoracic Society
  • Cancer Research UK
  • National Lung Cancer Forum for Nurses
  • Royal College of Nursing
  • Royal College of Pathologists
  • Royal College of Physicians

III Other consultees:

  • Department of Health
  • Greater Manchester PCT Cluster
  • Southampton, Hampshire, Isle of Wight and Portsmouth PCT Cluster
  • Welsh Government

IV Commentator organisations (did not provide written evidence and without the right of appeal):

  • Commissioning Support Appraisals Service
  • Department of Health, Social Services and Public Safety for Northern Ireland
  • Healthcare Improvement Scotland
  • British Thoracic Oncology Group
  • NHS Centre for Reviews and Dissemination and Centre for Health Economics, York
  • National Institute for Health Research Health Technology Assessment Programme
  • National Collaborating Centre for Cancer

C The following individuals were selected from clinical specialist and patient expert nominations from the consultees and commentators. They gave their expert personal view on crizotinib by attending the initial Committee discussion and providing written evidence to the Committee. They are invited to comment on the ACD.

  • Dr Fiona Blackhall, Consultant Medical Oncologist, nominated by Pfizer – clinical specialist
  • Dr Sanjay Popat, Consultant Medical Oncologist, nominated by Royal College of Physicians and Pfizer  – clinical specialist
  • Dr Jesme Fox, nominated by Roy Castle Lung Cancer Foundation – patient expert

D Representatives from the following manufacturer/sponsor attended Committee meetings. They contributed only when asked by the Committee chair to clarify specific issues and comment on factual accuracy.

  • Pfizer

This page was last updated: 18 April 2013

Accessibility | Cymraeg | Freedom of information | Vision Impaired | Contact Us | Glossary | Data protection | Copyright | Disclaimer | Terms and conditions

Copyright 2014 National Institute for Health and Care Excellence. All rights reserved.

Selected, reliable information for health and social care in one place

Accessibility | Cymraeg | Freedom of information | Vision Impaired | Contact Us | Glossary | Data protection | Copyright | Disclaimer | Terms and conditions

Copyright 2014 National Institute for Health and Care Excellence. All rights reserved.

Accessibility | Cymraeg | Freedom of information | Vision Impaired | Contact Us | Glossary | Data protection | Copyright | Disclaimer | Terms and conditions

Copyright 2014 National Institute for Health and Care Excellence. All rights reserved.