Addendum to Lilly submission and audit trail Workbook Alimta UK adaptation xlsm

Transcription

1 Addendum to Lilly submission and audit trail Workbook Alimta UK adaptation xlsm Changes that have affected base case results or sensitivity analyses: Worksheet jmen_resource Cells F17-F21, F29-F33 formula amended to reflect correct adverse event name (e.g. =(1-EXP(-E18))*cNausea ) due to name referencing error (previously all cells had referenced cneutro), the formula was also amended to change the rate per person cycle into a probability per cycle. Cells E29-E32 formula amended to reflect BSC pre-progression person cycles due to cell referencing error (previously reflected PEM pre-progression person cycles). Base case ICER changes from 49,258 to 49,278 Worksheet resource Cells E332 amended to 80 from 85 to reflect PARAMOUNT data (footnote also amended) due to data input error. This change then effects C305 and F305. Base case ICER changes from 49,278 to 49,679 Worksheet pem and bsc Columns DH-DI. The costs of docetaxel and erlotinib were missing from total cost calculations for JMEN resource use scenario (model dated columns DR, DT and DW, DX). Base case ICER changes from 49,679 to 47,576 Changes that have not affected default lifetime base case results or sensitivity analyses: Worksheets pem and bsc (Markov trace) Columns BE and BQ. Formula amendment to prevent negative numbers in extreme scenarios (e.g. formula amended to =IF((BD12+BE11-BT12)>0,BD12+BE11- BT12,0) ) Simplification of Markov trace layout for resource use (removed separate columns for JMEN cost data - columns DK-DO in the model dated 11/10/12). Checker cell included to cross validate resource use estimates (Column DS) Worksheets results2 and results3 NICE submission results tables added to automate results generation Macros created in VBA to run NICE submission results Worksheets parameters Cells E163-I165 - change in location of variables propctscans, propconsults, pchemohome (previously located in resource worksheet cells C172, C173 and C140), added into parameters sheet to automate sensitivity analyses for NICE

2 submission results. Note column E directly feeds into Markov trace. Columns F-I are used to restore default analyses. Worksheet resource Cells C172, C173 and C140 describe these resource use items and have been left in the model for reference purposes only. They no longer directly input in the Markov trace. Cells E174. Addition of analysis time into parameters sheet to automate sensitivity analysis on time horizon for lifetime analysis (e.g. 99% rather than 99.9% of patients who had died). Word document: Alimta NICE submission_merged_15 th October 2012_final.doc The following changes have been made to the original submission as a result of the changes to the model dated as detailed in the change audit on previous page. Note: page numbers refer to initial submission document. Executive summary We have reproduced the entire text of the executive summary for comprehensive review, however, only the cost-effectiveness results detailed below have been changed, and are highlighted in pink: page 15: deterministic and probabilistic ICERs, probabilities of being cost-effective at 50,000 WTP threshold page 16, Table 1: Base-case cost-effectiveness results 7.3 Clinical parameters and variables Section 7.3.6, page 109, Table 26 AE rates per cycle Placebo/BSC (fatigue and anaemia) 7.5 Resource identification, measurement and evaluation Section 7.5.7, page 130, Table costs updated due to input error in original submission (no change to model) Section 7.5.7, page 130, Table 46 AE rates and total costs for PARAMOUNT (Adapted from TA190; AE data PARAMOUNT), columns Rate per 21-day cycle (placebo - fatigue and anaemia) and Cost per patient cycle Paramount Section 7.5.7, page 131, Table 48 value per cycle for Treatment with AEs (preprogression with Pem/BSC, pre-progression with placebo/bsc) 7.6 Sensitivity analysis Section 7.6.2, page 135 and 136, Table 49 AE costs Section 7.6.2, page 136 Table 49 Relabelled HR OS and PFS as treatment effect OS and PFS 7.7 Results Section 7.7.4, page 141, Table 51 Model outputs by clinical outcomes, column Total Cost

3 Section 7.7.5, page 142, Table 52 Summary of QALY gain by health state, absolute increment column Section 7.7.5, page 142, Table 53 Summary of costs by health state, all columns Section 7.7.5, page 143, Table 55 Summary of predicted resource use by category of cost, all columns Section 7.7.6, page 143, Table 56 Deterministic basecase results, and associated text Section 7.7.7, pages , Table 57 Deterministic sensitivity analysis, and associated text on page 150. Final sensitivity analysis deleted because full set of results provided for adjusted analysis in response to clarification question B1b. Section 7.7.7, pages 147, Table 57 Deterministic sensitivity analysis. Relabelled HR OS and PFS as treatment effect OS and PFS Section 7.7.7, page 149, Table 58 Sensitivity analysis: Alternative parametric distributions, and associated text on page 150 Section 7.7.8, page 151, Table 59, Figure 16 Section page 152, Figure 17 Section , page 153, deterministic and probabilistic ICERs, probabilities of being cost-effective at different thresholds. 8 Assessment of factors relevant to the NHS and other parties Section 8.5, page 164, Incremental costs for treating AEs for patients receiving pemetrexed were updated to 6.60 per cycle to reflect the changes made to AE costs in the jmen_resource worksheet of the model as described above Section 8.5, page 164 and 165, Table 66 Adverse event cost modified, total cost per cycle and total cost per patient modified based on change in AE cost Section 8.7, page 167, The estimated budget impact in the first 5 years following recommendation has been updated to include the change in AE cost Section 8.7, page 167, Table 72 Cost with pemetrexed and net budget impact updated include change in AE cost

4 Revised text, tables and figures for submission following changes to unadjusted economic model Executive summary Please provide an executive summary that summarises the key sections of the submission. All statements should be directly relevant to the decision problem, be evidence-based when possible and clearly reference the relevant section of the submission. The summary should cover the following items. Executive summary Lung cancer Improving survival in lung cancer patients is a key government priority as shown by its inclusion in the NHS Outcomes Framework 2011/2012. Lung cancer is the leading cause of cancer-related death in England and Wales (Cancer Research UK 2010), with about a third of patients dying within one year after diagnosis. Survival in lung cancer patients is the worst among the big four cancers (lung, breast, bowel and prostate). Survival rates for lung cancer in both men and women have improved over the last two decades (Cancer Research UK). One-year survival in men increased from 15% in the 1970s to 29% in year and 10-year survival rates increased too though at a slower pace. These rates possibly do not reflect the technological advances that occurred later in the mid to late-2000s, i.e., availability of pharmacological treatment options like pemetrexed, and the biological agents erlotinib and gefitinib for use in the first-line setting which have since transformed the standard of care for patients with NSCLC in the UK. Treatment of lung cancer NSCLC is asymptomatic in the early stages of the disease. Since lung cancer is largely asymptomatic in the early stages, patients usually present at an advanced stage, by which time their cancer is likely to be inoperable. For those with non-resectable cancer, the treatment options are chemotherapy and radiotherapy. First-line chemotherapy treatment is given following diagnosis with the aim of reducing tumour size (response), improving progression-free and overall survival whilst maintaining healthrelated quality of life (HRQL). Maintenance chemotherapy treatment aims to prolong the response achieved in the first-line treatment setting in patients whose disease has not progressed, i.e. to extend the duration of disease control thereby maintaining HRQL, improving progression-free survival and overall survival with minimal side-effects. Second-line treatment aims to relieve symptoms due to disease progression. Maintenance treatment in the NHS The administration of an active maintenance treatment immediately following first-line therapy improves overall survival in NSCLC by allowing more patients to benefit from additional treatment while tumour and symptom burden is low, patient tolerance is high and before the inevitable deterioration in performance status and disease progression occurs. Market research data show that currently in the NHS only 6% of non-squamous NSCLC patients who receive first-line treatment go on to receive maintenance treatment. Although these figures appear low, they reflect the fact that active maintenance treatment for NSCLC is a relatively new concept with the first active treatment (pemetrexed for switch maintenance) being licensed for use in the NHS as recently as 2009, and has yet to become embedded within clinical practice in the NHS.

5 Pemetrexed in maintenance treatment of non-squamous NSCLC Pemetrexed was previously licensed for the treatment of locally advanced, metastatic (stage IIIB/IV) non-squamous NSCLC in the first-line and second-line settings and for maintenance treatment in patients without disease progression following first-line therapy with non-pemetrexed containing regimens (i.e., switch maintenance ). The recent (24 th October 2011) amendment to the licence allows the use of pemetrexed in advanced, metastatic (stage IIIB/IV) non-squamous NSCLC in patients who have not progressed after four cycles of pemetrexed/cisplatin first-line treatment. There are two licensed maintenance treatment options available in the NHS, pemetrexed and erlotinib. Pemetrexed has been recommended by NICE for switch maintenance treatment of nonsquamous NSCLC (TA 190). Erlotinib is not recommended by NICE for maintenance treatment of NSCLC. Pemetrexed continuation maintenance Pemetrexed is now licensed for treatment of patients with locally advanced or metastatic, nonsquamous NSCLC who have not progressed following four cycles of pemetrexed/cisplatin first-line. The current submission presents the clinical and cost-effectiveness case for pemetrexed continuation maintenance in this population. The PARAMOUNT study Since pemetrexed/cisplatin is the standard of care for first-line non-squamous NSCLC in the NHS, there was a clinical demand to determine whether patients receiving pemetrexed/cisplatin first-line would benefit from further treatment with pemetrexed monotherapy in the maintenance setting. The PARAMOUNT study was designed to address this question. Key results from this study are as follows: Pemetrexed-treated patients experienced a significantly higher PFS benefit of 1.28 months over placebo/bsc (median overall PFS at final data lock of 4.44 months compared to 2.76 months) and a 40% reduction in risk of disease progression (HR for pemetrexed/bsc vs placebo/bsc: 0.60). Pemetrexed-treated patients experienced significant OS benefit of 2.85 months (median overall survival of months vs months for pemetrexed/bsc vs placebo/bsc, log-rank p=0.0195) and a 22% reduction in the risk of death compared to placebo-treated patients (HR for pemetrexed/bsc vs placebo/bsc: 0.78). This OS benefit was in addition to that experienced by patients treated with pemetrexed/cisplatin in the first-line treatment setting. 1-year and 2-year survival rates for pemetrexed treated patients were 58% and 32% respectively, compared to 45% and 21% for placebo treated patients. EQ-5D data were collected in PARAMOUNT with compliance rates of over 80% in the maintenance phase. No statistically significant differences in changes from baseline in EQ-5D index scores were seen between pemetrexed/bsc and placebo/bsc. The analysis of performance status showed that patients were able to maintain their performance status and there were no between group differences in changes in performance status from baseline. These data show that patients can tolerate long-term pemetrexed continuation maintenance without significant detrimental impact on QoL. Pemetrexed was well-tolerated in PARAMOUNT, with an adverse events profile consistent with the known safety profile of pemetrexed given as single-agent switch maintenance treatment in the JMEN study and second-line treatment in the JMEI study. The grade 3/4 toxicities that were significantly different between pemetrexed and placebo were neutropenia (5.8% vs 0%, p=0.0002), anaemia (6.4% vs 0.6% p=0.001) and fatigue (4.7% vs 1.1%, p=0.044).

6 Patient perspective on pemetrexed continuation maintenance The implications of the PARAMOUNT study for patients suffering from this terminal disease are as follows: Pemetrexed continuation maintenance makes it possible for clinicians to give patients the most effective treatment (pemetrexed/cisplatin) upfront, so that patients are able get the most benefit in terms of increased survival and symptom palliation (Scagliotti et al 2008). Patients can continue pemetrexed monotherapy enabling them to maintain benefit of first-line treatment and avoid cisplatin-associated toxicities like nausea, vomiting, ototoxicity and neurotoxicity as well as hospital stays for cisplatin-required hydration. Pemetrexed continuation maintenance improves the outlook for patients suffering from nonsquamous NSCLC by providing them with an opportunity for increased survival while maintaining their performance status and without significant detrimental impact on their quality of life. Since patients are fit enough to receive treatment, this could potentially improve their chances of receiving further chemotherapy in the second-line setting. Pemetrexed as a single-agent treatment requires a ten-minute infusion once every three weeks and can be administered in chemotherapy units or in the community/at home. This has the added benefit of potentially moving care of these patients from the hospital into the community, which is more convenient for patients and carers. Pemetrexed continuation maintenance fulfils all the criteria of the NICE End of Life supplementary advice Pemetrexed continuation maintenance fulfils all three criteria specified in NICE s Supplementary Advice for Appraising Life Extending, End of Life Treatments and therefore the supplementary advice should be applied to this appraisal. Criterion 1: The cumulative population of patients eligible for pemetrexed treatment across all NSCLC indications and mesothelioma is 5,531 (4,034 NSCLC pts; 1,497 MPM pts), which is less than the population size implicitly set at < 7,000. Criterion 2: The overall survival benefit for patients treated with pemetrexed/bsc from the PARAMOUNT trial was 2.85 months. Due to the high censoring for OS data, an extrapolation of the trial survival data over a lifetime horizon was undertaken. This provided a modelled mean overall survival of 4.2 months in the basecase analysis (range from the parametric distributions explored: 3.4 to 4.7 months). Criterion 3: The median overall survival in England and Wales is lower than 24 months Cost-effectiveness analysis The economic analysis compared the cost-effectiveness of pemetrexed/bsc with that of placebo ( watch and wait )/BSC as continuation maintenance in patients with locally advanced, metastatic NSCLC (stage IIIB/IV) who have not progressed following four cycles of first-line pemetrexed/cisplatin. A cost-effectiveness analysis has been conducted from the perspective of the NHS in England and Wales with a lifetime horizon. The analysis is based on a Markov model populated with individual patient data (IPD) from the PARAMOUNT study. The survival models developed from the IPD are extrapolated and incorporated into an Excel-based state-transition Markov model. The economic evaluation gives a deterministic ICER of 47,576 and a probabilistic ICER of 48,218. A wide range of one-way sensitivity analyses have been conducted which demonstrates consistent results across a range of alternative plausible data inputs. The Cost Effectiveness Acceptability Curve (CEAC) shows that at a 50,000 WTP threshold pemetrexed/bsc is cost-effective in 54% of simulations.

7 Conclusion Pemetrexed continuation maintenance offers patients who currently have no treatment options immediately following first-line treatment with pemetrexed/cisplatin, but who are appropriate candidates for active chemotherapy, a cost-effective treatment under conventional thresholds when the end of life criteria are applied. Table 1. Base-case cost-effectiveness results Pemetrexed Placebo Incremental results Therapy costs* 13, ,125 Adverse event costs Follow up care costs 10,177 11, Terminal care costs 2,699 2, Total costs 26,064 13,912 12,153 LYG QALYs ICER 47,576 Note: * Therapy costs includes drug acquisition, delivery and additional monitoring costs (See table 5, Section for further details of cost categories);; LYG, life years gained; QALY(s), quality-adjusted life year(s); ICER, incremental cost-effectiveness ratio

8 Table 26. Summary of variables applied in the economic model Variable Value CI (distribution) Reference to section in submission Age, median (range) 61 (32-83) years N/A Section Gender 58% male; 42% female N/A Section m 2 based on mean UK Body Surface Area (BSA) m 2 BSA values weighted by gender from PARAMOUNT Clinical Outcomes: Incremental 4.21 months mean OS from randomisation. Overall survival Extrapolated using gamma distribution in the basecase analysis. Incremental 3.25 months mean PFS from randomisation. PFS Extrapolated using gamma distribution in the basecase analysis. Treatment discontinuation AE rates per cycle PEM/BSC AE rates per cycle Placebo/BSC Second-line chemotherapy use Utility values: Pre-progression >6 cycles prior to death Pre-progr >4 < 6 cycles prior to death Pre-progr > 2 < 4 cycles prior to death Pre-progr 0-2 cycles prior to death Post-progression >6 cycles prior to death Post-progr >4 < 6 cycles prior to death Post-progr > 2 < 4 cycles prior to death Post-progr 0-2 cycles prior to death 5.09 cycles placebo and 7.95 cycles pemetrexed. Not extrapolated in the basecase analysis. Neutropenia Nausea & Vomiting Fatigue Anaemia Neutropenia Nausea & Vomiting Fatigue Anaemia Pem/BSC 64% Placebo/BSC 72% Normal distribution of BSA by gender Probabilistic estimates have been derived for KM data using the standard error associated with the observed failure rate per cycle. (Beta distribution) In the extrapolated period multivariable regression functions generated using KM data have been entered in the model along with Cholesky decompositions that account for correlations between parameters Section 7.5 Section 6.5 Section 6.5 Section 6.5 (cycle data) No CI available Section No CI available Section Pem/BSC 58-68% Placebo/BSC 65-78% (Beta distribution) Utility values derived from a mixed regression model. Multivariable regression functions generated using PARAMOUNT individual patient data have been entered in the model along with a Cholesky decomposition to account for correlated parameters. (CI for the coefficients derived from the regression model are provided in Table 28) Section Section 7.4.9

9 Table 45. Unit costs for common grade 3/4 adverse events in PARAMOUNT (Costs inflated from those used in TA190) Grade 3-4 adverse events Reported Cost Year Cost per event (2009) Cost per event inflated to 2011 Neutropenia Nausea and vomiting Fatigue Anaemia Table 46. AE rates and total costs for PARAMOUNT (Adapted from TA190; AE data PARAMOUNT) Grade 3/4 AEs Placebo/BSC (n=180) Number of patients experiencing grade 3/4 AE AE rate % Rate per 21- day cycle Cost per patient cycle PARAMOUNT Neutropenia % Nausea / Vomiting % Fatigue % Anaemia % Total cost for placebo/bsc 0.83 Pem/BSC (n=359) Neutropenia % Nausea / Vomiting % Fatigue % Anaemia % Total cost for pem/bsc 7.43

10 Table 48. List of health states and associated costs in the economic model Health states Pre-progression with pem/bsc Pre-progression with placebo/bsc Post-progression with second-line chemotherapy Note: ** These costs are only applied in those cycles when second-line chemotherapy is administered. Post-progression with no secondline chemotherapy Costs applied Value (per cycle) (See Section 7.5.5) Pemetrexed acquisition 1,440 Pemetrexed delivery Pemetrexed monitoring (consultant and CT scans) BSC with active chemotherapy Treatment of AEs 7.43 BSC without active chemotherapy Treatment of AEs 0.83 Docetaxel acquisition** 1, Erlotinib acquisition** Docetaxel delivery** Erlotinib delivery** BSC with active chemotherapy** BSC without active chemotherapy, applied to remaining cycles after second-line chemotherapy has been discontinued Terminal care, applied to the final cycle before death 2, BSC without active chemotherapy Terminal care, applied to the final cycle before death 2, Death N/A N/A Table 49. Sensitivity analysis of parameter values and structural assumptions Parameter Basecase value Range or alternative value(s) Rationale and alternative data source AE costs AE and BSC costs Treatment effect for OS and PFS 7.43 pem/bsc 0.83 placebo/bsc AE costs: 7.43 pem/bsc 0.83 placebo/bsc BSC costs: 36.22/cycle if on active chemo 72.44/cycle if not on active chemo Treatment effect derived from K-M data +/- 10% Assumption BSC drug costs applied to every cycle plus hospitalisation, blood transfusion data and palliative radiotherapy rates and associated NHS reference costs applied. Upper and lower 95% CI In the basecase analysis the methods for costing AE and BSC resources are consistent with TA190, with costs inflated to The cost of BSC when not on active chemotherapy is assumed to include the cost of palliative radiotherapy. Instead of AE costs we have used hospitalisations and blood transfusion data from PARAMOUNT. Instead of BSC costs we have used BSC drug use and palliative radiotherapy data from PARAMOUNT. See Appendix 20 for full details. PARAMOUNT data

11 Table 51. Model outputs by clinical outcomes Outcome LYs QALYs Total Cost ( ) Pemetrexed ,064 Overall survival Placebo ,912 Incremental ,153 Pemetrexed ,592 Pre-progression Placebo Incremental ,123 Pemetrexed ,472 Post-progression Placebo ,443 Incremental Note: Numbers may not compute due to rounding; LY, life years; QALY, quality-adjusted life year Table 52. Summary of QALY gain by health state Health state QALY pemetrexed QALY placebo Increment % absolute increment Pre-progression % Post-progression % Total % Note: Numbers may not compute due to rounding; QALY, quality-adjusted life year. Adapted from Pharmaceutical Benefits Advisory Committee (2008) Guidelines for preparing submissions to the Pharmaceutical Benefits Advisory Committee (Version 4.3). Canberra: Pharmaceutical Benefits Advisory Committee Table 53. Summary of costs by health state Health state Cost pemetrexed Cost placebo Increment % absolute increment Pre-progression 13, , % Postprogression 12,472 13, % Total 26,064 13,912 12,153 87% Note: Numbers may not compute due to rounding. Adapted from Pharmaceutical Benefits Advisory Committee (2008) Guidelines for preparing submissions to the Pharmaceutical Benefits Advisory Committee (Version 4.3). Canberra: Pharmaceutical Benefits Advisory Committee

12 Table 55. Summary of predicted resource use by category of cost Item Cost pemetrexed Cost placebo Increment % absolute increment Therapy cost 13, ,125 - Adverse event cost Follow up care costs Terminal care costs % 10,177 11, % 2,699 2, % Total 26,064 13,912 12,153 87% Base-case analysis Please present your results in the following table. List interventions and comparator(s) from least to most expensive and present ICERs in comparison with baseline (usually standard care) and then incremental analysis ranking technologies in terms of dominance and extended dominance. The results of the deterministic basecase analysis are shown in Table 56. The deterministic results show that the mean overall survival gain is 0.35 LYG, i.e., 4.2 months, with a QALY gain of 0.26 for pem/bsc compared to standard care, i.e. placebo/bsc. The estimated ICER for pemetrexed/bsc compared to standard of care is 47,576 per QALY gained. The ICER is driven by the comparator being watch and wait with low associated costs compared to active intervention with pemetrexed continuation maintenance treatment. Table 56. Deterministic basecase results Technologies Total costs ( ) Total LYG Total QALYs Incremental costs ( ) Incremental LYG Incremental QALYs Incremental cost/lyg ICER ( ) incremental (QALYs) Placebo/BSC 13, Pem/BSC 26, , ,613 47,576 Note: Numbers may not compute due to rounding; ICER, incremental cost-effectiveness ratio; LYG, life years gained; QALYs, quality-adjusted life years

13 Table 57. Deterministic sensitivity analysis Parameter Pemetrexed costs Value or assumption in base case Incr. cost ( ) Incr. benefit (QALY) ICER Base case 12, ,576 Wastage excluded (assumes vial sharing) Wastage included 12, ,313 Concomitant vitamins and corticosteroid included Excluded 12, ,672 DH HRG daycase procurement costs for pemetrexed average 1,293 1,440 10, ,035 DH HRG daycase procurement costs for pemetrexed lower quartile 928 1,440 8, ,760 DH HRG daycase procurement costs for pemetrexed Upper quartile 1,611 1,440 13, ,858 Pemetrexed delivery cost Lower quartile 131 Daycase , ,201 Pemetrexed delivery cost Upper quartile 233 Daycase , ,352 Pemetrexed delivery cost Outpatient average 231 Daycase , ,290 Pemetrexed delivery cost Outpatient lower quartile 114 Daycase , ,676 Pemetrexed delivery cost Outpatient upper quartile 279 Daycase , ,773 Deliver 20% of pemetrexed in community or home setting at 96/cycle Daycase , ,918 Deliver 50% of pemetrexed in community or home setting at 96/cycle Daycase , ,930 ( ) Increase AE costs by 10%: 8.17 pem/bsc; 0.91 placebo/bsc Decrease AE costs by 10%: 6.68 pem/bsc; 0.75 placebo/bsc BSA based on PARAMOUNT IPD IPD: equivalent to mean 1.8m 2 ) (including wastage) (Each patient in cohort costed based on BSA based on PARAMOUNT IPD (Each patient in cohort costed based on IPD: equivalent to mean 1.8m 2 ) (excluding wastage) 7.43 pem/bsc 0.83 placebo/bsc 7.43 pem/bsc 0.83 placebo/bsc Mean BSA 1.79m 2 Wastage included Mean BSA 1.79m 2 Wastage included 12, ,599 12, ,553 12, ,586 12, ,430

14 Additional monitoring for patients on pemetrexed: Every 12 weeks: 1 extra consultant visit for all patients and 1 extra CT scan for 3% of cohort Every 12 weeks: 1 extra consultant visit for all patients and 1 extra CT scan for 20% of cohort Every 12 weeks: 1 extra consultant visit for all patients and 1 extra CT scan for 50% of cohort Every 12 weeks: 1 extra consultant visit for all patients and 1 extra CT scan for 100% of cohort Every 6 weeks: 1 extra consultant visit for all patients and 1 extra CT scan for 3% of cohort on pemetrexed Every 6 weeks: 1 extra consultant visit for all patients and 1 extra CT scan for 20% of cohort on pemetrexed Every 6 weeks: 1 extra consultant visit for all patients and 1 extra CT scan for 50% of cohort on pemetrexed Every 6 weeks: 1 extra consultant visit for all patients and 1 extra CT scan for 100% of cohort on pemetrexed Second-line chemotherapy costs Docetaxel average price from DH CMU emit database (accessed ) Erlotinib & docetaxel equivalent to average docetaxel emit price DH HRG daycase procurement costs for erlotinib and docetaxel average Erl: 2,165; Doc 832 Cycle data from PARAMOUNT: 3.26 Docetaxel 5.25 Erlotinib BSC and terminal care costs No differential BSC costs applied according to active vs no active treatment 36.22/cycle Over 24-week period: 1 extra consultant visit for all patients & 1 extra CT scan for 3% of cohort 12, ,056 12, ,243 12, ,574 12, ,126 12, ,016 12, ,391 12, ,053 13, ,157 1,023 12, ,109 Erlotinib Docetaxel 1,023 Erlotinib Docetaxel 1, Docetaxel 6.27 Erlotinib 36.22/cycle if on active chemo 72.44/cycle if not on active chemo 12, ,742 11, ,050 12, ,348 12, ,222

15 No differential BSC costs applied according to active vs no active treatment 72.44/cycle No BSC applied (terminal cost applied) No terminal or BSC costs applied PARAMOUNT resource use data BSC drug costs, hospitalisation, blood transfusion and palliative radiotherapy rates and associated NHS reference costs. Second-line chemotherapy cycles based on PARAMOUNT data Utilities Assume no treatment effect associated with pemetrexed treatment during maintenance treatment (i.e. preprogression); utilities equivalent to BSC pre-progression Utility values from TA190 (Scenario 5 JMEN values) 36.22/cycle if on active chemo 72.44/cycle if not on active chemo 36.22/cycle if on active chemo 72.44/cycle if not on active chemo 36.22/cycle if on active chemo 72.44/cycle if not on active chemo 2,825 terminal care costs AE costs: 6.51 pem/bsc 0.29 placebo/bsc BSC costs: 36.22/cycle if on active chemo 72.44/cycle if not on active chemo Second-line chemo cycles: Doc:4.82 ; Erl: 6.27 Apply non-significant disutility ( ) PARAMOUNT EQ- 5D data 12, ,088 12, ,356 12, ,509 13, ,279 12, ,447 12, ,977

16 Efficacy Post-trial treatment effect: pem/bsc is equivalent to placebo/bsc, i.e., treatment benefit for trial period only Treatment effect assumed to continue beyond trial duration 12, ,012 OS Treatment effect 95% lower CI (Time ratio: 1.02) OS treatment effect 12, ,957 OS Treatment effect 95% upper CI (Time ratio: 1.46) OS treatment effect 12, ,517 PFS Treatment effect 95% lower CI (Time ratio: 1.23) PFS treatment effect 12, ,587 PFS Treatment effect 95% upper CI (Time ratio: 1.76) PFS treatment effect 12, ,563 Structural Discounting costs at 0% 3.5% 12, ,474 Discounting health effects at 0% 3.5% 12, ,506 Discounting costs and effects at 0% 3.5% 12, ,327 Discounting costs at 6% 3.5% 12, ,999 Discounting health effects at 6% 3.5% 12, ,423 Discounting costs and effects at 6% 3.5% 12, ,811 Time horizon - 6 years (i.e. stop Markov trace at 105 cycles) years 12, ,345 Time horizon - 10 years (i.e. stop Markov trace at 174 cycles) years 12, ,502 Time horizon 8.97 years (i.e. stop Markov trace at 156 cycles) i.e. lifetime horizon based on when 99% of patients have died Cut-points for extrapolation OS: Approx 15% at risk by arm: i.e. 34 cycles placebo/bsc (16%) & 38 cycles pem/bsc (15.9%) OS: Approx 25% at risk by arm: i.e. 29 cycles placebo/bsc (25.6%) & 33 cycles pem/bsc (26.7%) OS: using all available observed OS data i.e. 49 cycles placebo/bsc & 50 cycles pem/bsc years (99.9% patients died) 12, ,979 20% at risk 12, ,471 20% at risk 12, ,675 20% at risk 12, ,272 Fully parametric OS with observed PFS & treatment discontinuation 20% OS at risk 12, ,091

17 Table 58. Sensitivity analysis: Alternative parametric distributions Alternative parametric distribution Parameter value or assumption in basecase Inc. cost ( ) Inc. mean OS (months) Deterministic Inc. benefit (QALY) ICER ( ) Inc. cost ( ) Inc. mean OS (months) Probabilistic Inc. benefit (QALY) Base case 12, ,576 12, ,866 Exponential Gamma 12, ,133 12, ,565 Weibull Gamma 12, ,187 12, ,458 Gompertz Gamma 12, ,742 12, ,840 Log-normal Gamma 12, ,532 12, ,335 Log-logistic Gamma 12, ,902 12, ,465 ICER ( )

18 Overall, the majority of the results from the sensitivity analyses range from 47,000 to 50,000. This shows the model is robust and provides a high level of consistency across a wide range of alternative plausible one-way analyses. Looking at the maximum and minimum values from the sensitivity analysis the results show that the deterministic ICERs range from 31,760 to 58,840. The 31,760 ICER is based upon the use of the lower quartile NHS reference cost for daycase procurement of pemetrexed. As discussed in Section 7.5, the procurement costs were not considered to be appropriate for costing pemetrexed in the basecase analysis due to the wide variation in chemotherapy regimens covered by the relevant NHS code and the resultant insensitivity of the associated cost of an individual drug. If this implausible result is excluded, the remaining ICERs range from 40,517 to 58,840. The latter is the result of adopting a Gompertz parametric distribution which was considered to have a poor fit based on both internal and external validity. The lower value results from using the 95% upper CI for OS treatment effect, i.e. the time ratio for basecase gamma distribution. The key drivers of the model are: Efficacy of pemetrexed: including both the implementation of the upper and lower confidence intervals for the OS treatment effect and changing the assumption of the post trial treatment effect. Use of alternative parametric distributions; Use of utility values from external literature as used in TA190. The utility values used in TA190 were derived from studies in NSCLC patients being treated with second-line chemotherapy which is likely to have lower face validity than using utility data directly derived from patients with the condition being assessed. Other variables in the sensitivity analyses have a lower impact on the ICER. For example, Increasing the frequency of monitoring and proportion of patients receiving CT scans from current levels of UK clinical practice within anticipated clinical practice scenarios does not impact the ICER greatly. Only when monitoring is modelled to occur every 6 weeks with all patients receiving a CT scan does the ICER increase to 51,157. The use of alternative chemotherapy delivery costs for pemetrexed in an outpatient or community/home setting do not impact the ICER significantly, however, delivery in a community or home setting may be preferred by some patients and free up capacity in chemotherapy units.

19 Table 59. Probabilistic basecase results Technologies Total costs ( ) Total LYG Total QALYs Placebo/BSC 13, Incremental costs ( ) Incremental LYG Incremental QALYs Incremental cost/lyg ICER ( ) incremental (QALYs) Pem/BSC 26, , ,995 48,218 Note: Numbers may not compute due to rounding ICER, incremental cost-effectiveness ratio; LYG, life years gained; QALYs, quality-adjusted life years Figure 16. Incremental cost-effectiveness plane for basecase analysis Figure 17. Cost-effectiveness acceptability curve What were the main findings of each of the sensitivity analysis? The economic evaluation of pemetrexed continuation maintenance compared to placebo in patients with advanced NS NSCLC gives a deterministic ICER of 47,576 and a probabilistic ICER of 48,218 (see Tables 56 and 59). A wide range of one-way sensitivity analyses have been conducted which demonstrates consistent results across a range of alternative plausible

20 data inputs. See Section The results from both the deterministic and the probabilistic analyses are in the similar range showing consistency. The CEAC shows that at a 50,000 WTP threshold pemetrexed/bsc is cost-effective in 54% of simulations. 8. Assessment of factors relevant to the NHS and other parties 8.5. Although very few grade 3/4 AEs occurred in either arm of the trial, pemetrexed was associated with a higher rate of AEs than the BSC arm. Incremental costs for treating AEs for patients receiving pemetrexed are 6.60 per cycle, (see Table 46, Section 7.5.7). Table 66. Summary of pemetrexed costs included in the budget impact analysis Costs Pemetrexed Chemotherapy 1, Administration Adverse event costs 6.60 Monitoring BSC and terminal care Excluded Total cost/cycle 1, Mean no. of cycles 7.9 Total cost/patient 13, What is the estimated annual budget impact for the NHS in England and Wales? We have assumed a relatively modest market share in the first year as the new treatment paradigm becomes established. The estimated annual budget impact in the first 5 years following recommendation for use in the NHS in England and Wales ranges from 468,225 in 2013 to 2,606,872 in The estimated budget impact is shown in Table 72. Table 72. Annual budget impact for pemetrexed in England and Wales in the first five years post-launch/nice recommendation No. of patients eligible for pemetrexed continuation maintenance Cost without pemetrexed maintenance therapy Market share of eligible patients ,592,075 2,592,075 2,592,075 2,592,075 2,592,075 7% 17% 34% 38.5% 38.5% No. pemetrexed patients Cost with pemetrexed 3,060,300 3,756,309 4,907,888 5,198,947 5,198,947 Net Budget Impact 468,225 1,164,234 2,315,813 2,606,872 2,606,872 Note: Small discrepancies in values are due to rounding, calculations were performed in Excel.