Ivacaftor for the treatment of patients with cystic fibrosis and the G551D mutation: a systematic review and cost-effectiveness analysis

Identification of studies

Systematic searches were undertaken to locate randomised controlled trials (RCTs) assessing ivacaftor. Searches were not limited by date, language or publication status (unpublished or published). The following databases were searched from inception in May 2012 with searches updated in July 2012:

• MEDLINE (OvidSP) 1946–April 2012 week 4

• MEDLINE In-Process & Other Non-Indexed Citations (OvidSP) up to 2 May 2012

• MEDLINE Daily Update (OvidSP) up to 2 May 2012

• EMBASE (OvidSP) 1974–2012 week 17

• Latin American and Caribbean Health Sciences Literature (LILACS) (Biblioteca Regional de Medicina) up to 4 May 2012

• Cochrane Database of Systematic Reviews (CDSR) (Wiley Online Library) up to Issue 4:2012

• Cochrane Central Register of Controlled Trials (CENTRAL) (Wiley Online Library) up to Issue 4:2012

• *Database of Abstracts of Reviews of Effects (DARE) (Wiley Online Library) up to Issue 2:2012 (CRD) up to 3 May 2012

• *NHS Economic Evaluation Database (NHS EED) (Wiley Online Library) up to Issue 2:2012 (CRD) up to 3 May 2012

• *Health Technology Assessment (HTA) Database (Wiley Online Library) up to Issue 2:2012 (CRD) up to 3 May 2012.

*For completeness DARE, NHS EED and HTA databases were searched through both the Wiley Online Library and the CRD host sites.

The EMBASE strategies were independently peer reviewed by a second information specialist, using the Peer Review of Electronic Search Strategies Evidence-Based Checklist (PRESS-EBC). 19 Supplementary searches were undertaken to identify unpublished and ongoing studies on the following resources:

• metaRegister of Controlled Trials (internet): www.controlled-trials.com

• National Institutes of Health Clinicaltrials.gov (internet): www.clinicaltrials.gov

• World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (internet): www.who.int/ictrp/en/.

The following conference proceedings were searched, where possible, from 2007 until the most recent conference (up to July 2012):

• European Cystic Fibrosis Society (ECFS) conference: www.ecfs.eu/conferences/main

• North American Cystic Fibrosis Conference (NACFC): www.nacfconference.org/

• International Congress on Pediatric Pulmonology (CIPP): www.cipp-meeting.org/index.htm.

The bibliographies of retrieved articles and relevant systematic reviews were screened for additional studies. Identified references were downloaded into EndNote bibliographic management software (Thomson Reuters, CA, USA) for de-duplication and then into Microsoft Access (Microsoft Corporation, Redmond, WA, USA) for further assessment and handling. Details of the search strategies can be found in Appendix 1.

Inclusion and exclusion criteria

Studies that fulfilled the following criteria were eligible for inclusion.

Data extraction strategy

Data were extracted using a standardised data extraction form by one reviewer and checked by another. Disagreements were resolved through discussion or referral to a third reviewer where necessary. Data were extracted on the primary outcome, lung function, and the following additional outcomes, where reported: mortality, weight, BMI, sweat chloride, respiratory symptoms, reduction in pulmonary exacerbations, exercise tolerance, adverse effects of treatment, HRQoL and utilisation of hospital resources. Data were extracted for 24-week follow-up (intermediate) and after the longest duration of follow-up reported. If data were available for different patient subgroups (e.g. age, disease severity, region) then data were extracted separately for each subgroup. If composite end points were reported, data were extracted on the definition of the end point, results, and, if sufficient data were available, the events that contributed to the end point. There were some discrepancies in data reported in different sources. In such situations data were extracted from a single source based on the following hierarchy: supplementary results report > journal article > conference abstract > manufacturer’s dossier > press release. Details on discrepancies in figures reported in the different reports are summarised in Appendix 5.

Critical appraisal strategy

Trials were assessed for methodological quality using the Cochrane risk of bias tool. 20 This includes items covering selection bias (random sequence generation and allocation concealment), performance bias (participant blinding), detection bias (blinding of outcome assessors), attrition bias (incomplete outcome data) and reporting bias (selective reporting of results). Each domain was assigned a rating of high, low, or unclear. Each trial was assigned an overall rating of the risk of bias. If at least one of the domains was rated as ‘high’ the trial was considered at high risk of bias and if all domains were judged as ‘low’ the trial was considered at low risk of bias; otherwise, the trial was considered at ‘unclear’ risk of bias. The risk of bias assessment was incorporated into the data extraction form and was conducted as part of the data extraction process.

Methods of data synthesis

We did not have sufficient data to conduct a formal meta-analysis. Data were tabulated and discussed in a narrative review. Where possible, results were grouped by age, lung function, disease severity and prior treatment (including consideration of intolerance to treatments). Dichotomous data were summarised as relative risks (RRs) or hazard ratios together with 95% confidence intervals (CIs). If continuous data were normally distributed then mean differences (MDs) and 95% CIs were calculated; otherwise, we reported the results of non-parametric statistical analyses conducted by the study authors. Where sufficient data were available, results were displayed graphically using forest plots. Publication bias was not formally assessed because of the very small number of trials included.

Quantity and quality of research available

The searches identified 256 references, of which 29 reports were considered potentially relevant and full-text copies were obtained. Three studies (16 reports) fulfilled the inclusion criteria: two Phase III RCTs and one open-label study (Figure 1). Three Phase II RCTs were excluded; all reported short-term outcomes only and one trial did not report any of the outcomes specified in the inclusion criteria. Details of these studies are summarised in Appendix 4.

FIGURE 1.

Flow of studies through the review. CA, conference abstract; JA, journal article; TR, trial registry entry.

Summary of included studies

The first RCT was conducted in adults (‘adults’ study’) and was published as a full-text report,21 with the study protocol and further results available as supplementary information from the journal website. Details were also reported in four conference proceedings22–25 and one trial registry entry. 26 The second RCT was conducted in children (‘children’s study’) and full results have not yet been reported. Details were available only from four conference abstracts,23,24,27,28 a press release29 and a trial registry entry. 30 An open-label extension study of the two included RCTs was also included. Details were available only from two conference abstracts,31,32 a press release29 and a trial registry entry. 33 Additional details on all three studies were submitted by Vertex Pharmaceuticals, the manufacturer of ivacaftor,16 and from FDA reports prepared to support the licensing recommendation. 34–36 Confidential information provided by the manufacturer has been removed throughout this report.

Baseline details of the two included RCTs are summarised in Table 1. The flow of patients through each study is summarised in Figure 2. Both studies were funded by Vertex Pharmaceuticals and were conducted in centres across the USA, Australia and Europe. Inclusion and exclusion criteria were similar with the exception of age: the study in adults enrolled adults and children aged ≥ 12 years; the study in children was restricted to children aged 6–11 years. All patients enrolled in the adults’ and children’s studies were eligible for inclusion in the open-label study. One patient who was in the placebo arm of the adults’ study did not enter the open-label study. Oral ivacaftor tablets were administered at a dose of 150 mg every 12 hours; the two RCTs also included an arm in which patients received matching placebo. The two RCTs were 48 weeks in duration. The open-label study was a further 96 weeks in duration with results currently available for 48 weeks’ follow-up (96 weeks’ ivacaftor treatment) in adults and 24 weeks’ follow-up (72 weeks’ ivacaftor treatment) in children.

FIGURE 2.

Flow of patients through the RCTs.

Risk of bias

The full results of the risk of bias assessment, including the support for judgement, are reported in Appendix 3. The rating of each bias criterion for each of the two RCTs is summarised in Table 2. The open-label study was not assessed for risk of bias, as this was a continuum of the two RCTs with all patients receiving ivacaftor and so issues relating to randomisation and blinding no longer applied. The study in adults was clearly reported and the availability of the study protocol meant that each of the risk of bias criteria could be assessed in detail. This study was rated as low risk of bias for all criteria. Fewer details were available for the study in children, as this has not yet been published in full. Randomisation, allocation concealment and incomplete outcome data were rated as unclear as there was insufficient information to make a judgement on these. The study was reported to have been double-blinded and so both blinding criteria were judged as low risk of bias. The study was also judged to be at low risk of bias for selective outcome reporting, as it appears that results for all relevant outcomes were reported.

Lung function

Both RCTs reported significantly greater changes from baseline in all measures of lung function in patients receiving ivacaftor compared with those receiving placebo at 24 and 48 weeks (Table 3 and Figure 3). The primary outcome was the absolute change from baseline in the percentage predicted FEV₁, expressed as a percentage of the predicted values for patients with similar height, age and sex. The MD between ivacaftor and placebo in ‘relative change from baseline in percentage predicted FEV₁’ and the MD in ‘actual FEV₁’ at 24 and 48 weeks were also assessed. The study in adults indicated that differences in lung function between the ivacaftor and placebo groups were due to improvements in lung function in the ivacaftor group while lung function in those in the placebo group stayed approximately the same or showed very slight decreases (Figure 4). (Commercial-in-confidence information has been removed.) The mean change from baseline in the adults’ and children’s trials and in the open-label extension at the various measurement points for each treatment group are shown in Figures 4 and 5. The improvement in lung function occurred very soon after treatment initiation with a MD between ivacaftor and placebo in change from baseline in percentage predicted FEV₁ of 9.17% after 2 weeks of treatment in the adults’ study. Results from the children’s study supported this initial improvement with a MD of 12.85% after 2 weeks of treatment (commercial-in-confidence information has been removed).

FIGURE 3.

Mean difference in change from baseline in percentage predicted FEV₁ (95% CI) in patients receiving ivacaftor compared with placebo in the adults’ study.

FIGURE 4.

Mean absolute change from baseline in percentage predicted FEV₁ in adults in the RCT and open-label study. (Commercial-in-confidence information has been removed.)

FIGURE 5.

Mean (95% CI) absolute change from baseline in percentage predicted FEV₁ in children in the RCT and open-label study. (Commercial-in-confidence information has been removed.)

The adults’ study also reported results stratified according to age, sex, study region and baseline lung function. Ivacaftor treatment resulted in significant improvements in absolute change in FEV₁ for all subgroups investigated (Table 4). CIs around estimates stratified according to subgroup were not reported and so it was not possible to formally investigate differences between subgroups. (Commercial-in-confidence information has been removed.) The children’s study did not report numerical results separately according to subgroups but results were presented graphically. This suggested that ivacaftor was associated with significantly greater improvements in absolute change in FEV₁ compared with placebo for the following subgroups: Europe, ≤ 90% predicted FEV₁ and girls. No significant differences were found for Australia, North America, > 90% predicted FEV₁ or for boys, although all point estimates favoured ivacaftor. The small number of children in each subgroup means that the study may have lacked power to detect significant differences in these subgroups.

Pulmonary exacerbations

Pulmonary exacerbations were defined in the adults’ study using modified Fuchs criteria38 of new or a change in antibiotic therapy [intravenous (i.v.), inhaled or oral] for any four or more of the following symptoms: new or increased haemoptysis; increased cough; increased dyspnoea; malaise, fatigue or lethargy; temperature above 38 ºC; anorexia or weight loss; sinus pain or tenderness; change in sinus discharge; change in physical examination of the chest; decrease in pulmonary function by 10%; radiographic change indicative of pulmonary infection. The number and severity of pulmonary exacerbations (patients with pulmonary exacerbation and total exacerbations) at both 24 and 48 weeks were significantly reduced in the ivacaftor group compared with placebo group in the adults’ study (Table 5 and Figure 6). The mean number of days with pulmonary exacerbations, mean number of days with exacerbations requiring i.v. antibiotics and number of days hospitalised with exacerbations were also significantly lower among the ivacaftor treatment group (Table 6). (Commercial-in-confidence information has been removed.) The authors of the study in children reported that exacerbations were uncommon in both groups.

FIGURE 6.

Relative risk (95% CI) of exacerbations in adults receiving ivacaftor compared with placebo (data are at 48 weeks’ follow-up unless otherwise stated).

Other outcomes

Other outcomes reported in the studies included change from baseline in ivacaftor and placebo groups for quality of life (QoL) [measured using the respiratory domain of the Cystic Fibrosis Questionnaire Revised (CFQ-R)], sweat chloride, weight, BMI and BMI-for-age z-score. Each of these outcomes was reported in both the adults’ and children’s studies at 24 and 48 weeks’ follow-up and quality of life and weight were also reported in the open-label study. There were significantly greater improvements in the ivacaftor group than in the placebo group for all outcomes at all time points with the exception of quality of life in children, which failed to reach statistical significance at either 24 or 48 weeks’ follow-up (Table 7, Figure 7). Patients who had received ivacaftor in the RCT continued to gain weight in the open-label study. (Commercial-in-confidence information has been removed.) The mean absolute change from baseline in CFQ-R respiratory domain scores in adults and children are summarised in Figures 8 and 9 respectively.

FIGURE 7.

Mean difference (95% CI) in change from baseline in (a) sweat chloride, (b) weight and (c) quality of life (measured using the CFQ-R) in patients receiving ivacaftor compared with placebo.

FIGURE 8.

Mean (95% CI) absolute change from baseline in CFQ-R respiratory domain score in adults in the RCT and open-label study. (Commercial-in-confidence information has been removed.)

FIGURE 9.

Mean (95% CI) absolute change from baseline in CFQ-R respiratory domain score in children in the RCT and open-label study. (Commercial-in-confidence information has been removed.)

Adverse events and withdrawals

Adverse events were mainly minor and comparable across treatment groups and studies (Table 8). The most commonly reported adverse events were pulmonary exacerbation, cough, headache, upper respiratory tract infection and oropharyngeal pain. Figure 10 shows the RR for each adverse event in intervention compared with control arms for the adults’ and children’s studies. In the adults’ study, there was a greater risk of a serious adverse event (RR 0.67, 95% CI 0.36 to 0.89), pulmonary exacerbation (RR 0.64, 95% CI 0.47 to 0.85), and decreased lung function test (RR 0.25, 95% CI 0.05 to 0.82) in the placebo group and a small increased risk of rash (RR 2.52, 95% CI 1.01 to 8.05) and dizziness (RR 9.40, 95% CI 1.51 to 56.39) associated with ivacaftor. However, these differences were not found in the children’s study, which reported RRs very close to 1 for each of these events. The children’s study did not find any significant differences between treatment groups.

FIGURE 10.

Relative risk (95% CI) of adverse events in ivacaftor compared with placebo for the (a) adults’ and (b) children’s studies. NR, not reported.

Both RCTs reported more overall withdrawals and withdrawals due to adverse events in the placebo group than in the ivacaftor group (Table 9). Two patients in the open-label trial discontinued treatment before the 12-week visit, one due to pregnancy and one due to the adverse event of suicidal depression.

Parameter estimates

Three of the 23 included studies contributed to the model. 8,21,43 Additionally, a more contemporary, as yet unpublished, version of the UK CF Registry database (2011) was used to inform cost parameters. 44 The adults’ study included in the review of clinical effectiveness was also included in the economic review. 21 UK-based generic utility values of CF patients by percentage predicted FEV₁ category were obtained from Gee et al. 45 and expressed in terms of SF-36. The prevalence rates of diabetes mellitus, Staphylococcus aureus infection and Burkholderia cepacia infection were derived from the UK CF Registry,8 which was also a data source for the proportion of genotyped patients and the proportion who were eligible for, and in receipt of, a lung transplant. Following correspondence with the Cystic Fibrosis Trust, audit data from the 2011 registry were obtained. These were analysed to provide information on costing, particularly in relation to tariff bands, expensive drugs and implantation of venous access devices.

No evidence was derived from the other included studies and the studies included as background. These studies were not useful for the model as, on detailed review, they did not:

1. provide useful evidence for use in a UK setting

2. transparently report costs or effects; or

3. provide utility values that were linked to percentage predicted FEV₁ bands.

Population

The starting patient population for this individual patient simulation is the population in the two RCTs included in the clinical effectiveness review (adults’ and children’s studies). 21,27 The analysis is therefore based on adults and children aged ≥ 6 years at the time of the start of the clinical trials in 2010. The cost-effectiveness study is conducted from a NHS perspective and so the reference population is the total CF population in England. There is therefore a potential concern regarding the generalisability of the model. To assess this, we compared baseline characteristics of patients included in the RCTs with details of patients included in the UK CF Registry, an anonymised database of all those with CF in the UK, maintained by the Cystic Fibrosis Trust (Table 10). 8 The median age of patients included in the UK CF Registry was lower than that of patients included in the trials, which is explained by the fact that patients aged < 6 years were excluded from the trials. The proportion male to female and median percentage predicted FEV₁ were comparable. Table 10 provides a summary of key characteristics of patients in the ivacaftor trials compared with UK CF Registry data. We did not make any modifications to the population in the model.

Strategies

To estimate the lifetime impact of ivacaftor in terms of costs and effects (QALYs) on CF patients, standard care (standard care strategy) was compared with ivacaftor plus standard care (ivacaftor + standard care strategy). Standard care consisted of CF-related medication [pancreatic enzymes, dornase alfa (DNase) (Pulmozyme^®, Roche), inhaled corticosteroids, bronchodilators, prednisone and antibiotics], devices (oxygen vests, nebulizers and other airway clearance and respiratory devices) and respiratory therapy. 16 We did not make any modifications to the strategies in the model.

Disease progression model

The model simulates the disease progression of CF patients included in the trials beyond the trial duration as an independent decline in percentage predicted FEV₁ (no change in any other characteristics, e.g. exacerbation rate). The probability of death is a function of the percentage predicted FEV₁, number of pulmonary exacerbations per year, infections with S. aureus (yes or no), infection with B. cepacia (yes or no), diabetes (yes or no), weight-for-age z-score and pancreatic sufficiency status (yes or no) (Table 11).

Improvement in the percentage predicted FEV₁, exacerbations and weight-for-age z-score associated with ivacaftor21 is translated into better survival of the patients. Each individual from the two RCTs with certain baseline characteristics runs through each treatment arm of the model. Every 3 months, patients’ characteristics are updated, based on efficacy outcomes and natural disease progression, and fed back into the model to estimate the survival of the patient. The estimated 3-monthly survival probability is then multiplied by the survival probability at the beginning of the 3-month period leading to cumulative survival probabilities. In addition, the HRQoL and health-care costs for the patients during the 3-month period are calculated. HRQoL values defined in terms of utility values were assumed to be dependent on the level of percentage predicted FEV₁ with a decrease in percentage predicted FEV₁ resulting in a decrease in utility. Costs were assumed to be dependent on percentage predicted FEV₁ and on age with a decrease in percentage predicted FEV₁ age resulting in an increase in costs. Adding up all the costs and effects generated in each time step leads to total costs and effects for each individual for both strategies. Average costs and QALYs per strategy are then used to estimate the incremental cost-effectiveness ratio (ICER).

Survival function

The probability of dying was estimated by means of a hazard function adapted from Liou et al. 47 depending on age, sex, percentage predicted FEV₁, number of pulmonary exacerbations, infections with S. aureus, infection with B.cepacia, diabetes, weight-for-age z-score and pancreatic sufficiency status (see Table 11). This study found no evidence of an association between other clinical parameters (e.g. height and infection with P. aeruginosa) and survival; these parameters were therefore not included in the survival function. Table 11 presents the original survival function based on Liou et al. ,47 the input estimates used by the manufacturer and our updated input estimates. The hazard function was estimated by subtracting the value of each individual patient characteristic from the reference values listed in Table 11.

The proportion female in the Liou et al. 47 study was 47% and was used as a reference value for the survival function. This reference value is compared with the sex status of patients included in the ivacaftor trials. Baseline values of percentage predicted FEV₁ are based on individual baseline estimates of the patients and compared with a reference baseline percentage predicted FEV₁ value of 67.7%. 47 The weight-for-age z-score was assumed to be constant over a lifetime period from the baseline score. Individual weight-for-age z-score estimates were used for the simulation of the disease progression and compared with a reference value of –0.85 based on Liou et al. 47 The number of exacerbations, based on trial data,1 was age dependent; patients aged ≥ 12 years treated with standard care were assumed to experience 1.4 exacerbations annually, while patients between 6 and 11 years were assumed to experience zero events annually. These estimations of the exacerbations are kept constant during the entire model duration and were compared with a reference value of 1.1 exacerbations per year. 47 The prevalence of diabetes mellitus, S. aureus infection and B. cepacia infection were not available from the trial data, and therefore age-specific percentages of patients with these conditions were derived from the UK CF Registry Annual Data Report8 and compared with the presented reference values. 47 Pancreatic insufficiency has a negative impact on the survival of patients and was therefore included in the survival function. Individual pancreatic insufficiency status is compared with the reference value. 47

Annual decline in percentage predicted forced expiratory volume in 1 second

The age-dependent annual decline in percentage predicted FEV₁ in the standard care arm was based on expert consultation. An annual decline in percentage predicted FEV₁ of 2% was used for patients aged between 10 and 20 years. For patients < 10 years or > 20 years the manufacturer assumed an annual decline of 1%.

Treatment effect

Based on the results of the clinical trials, treatment with ivacaftor was assumed to lead to an (almost) immediate improvement in the percentage predicted FEV₁, the decline in percentage predicted FEV₁ and the weight-for-age z-score, and a reduction in the annual number of exacerbations.

An initial age-dependent absolute improvement from baseline percentage predicted FEV₁ was applied to model the impact of ivacaftor on percentage predicted FEV₁ and possible reductions in the number of lung transplantations. This improvement (difference between ivacaftor group and placebo), based on the Phase III RCTs, was 10.5% in patients ≥ 12 years at treatment initiation21 and 10% for patients between 6 and 11 years. 27

For the long-term assessment of clinical effectiveness, extrapolation beyond observed data was required. The manufacturer assumed as base-case scenario that owing to the treatment no decline in percentage predicted FEV₁ would occur thereafter. Alternative efficacy scenarios for the rate of FEV₁ decline in the ivacaftor–standard care treatment arm were also investigated by the manufacturer (Table 14).

The trials21,27 showed an improvement in weight-for-age z-score. This improvement was age dependent and was estimated to be 0.33 for patients aged ≥ 12 years at treatment initiation. Children aged between 6 and 11 years at treatment initiation have an initial increase of 0.39. The initial increase remains over a lifetime period.

The annual reduction in the total number of exacerbations due to ivacaftor was estimated in the clinical trials to be 0.8 (RR 0.45). 21 This absolute reduction is used only for patients ≥ 12 years since the model assumed that patients younger than 12 do not experience exacerbations. The absolute decline of 0.8 exacerbations was kept constant over the whole model duration. In the manufacturer’s submission, ivacaftor was assumed not to influence the prevalence of diabetes or infections and therefore the same prevalence was used for the ivacaftor–standard care strategy.

Utilities

The manufacturer used utility values that were measured during the clinical trials. These utility values were obtained using baseline and end-of-trial (i.e. 48-week) EQ-5D scores obtained from patients in the trials. This generic measure of HRQoL was then adjusted by North American/European normative values to determine the utility scores. These utilities were linked to disease severity expressed in percentage predicted FEV₁. Table 15 presents the utility values by percentage predicted FEV₁ category used in the manufacturer’s model.

Costs of standard care

The annual costs of CF patients in the manufacturer’s model consist of two components: drug costs of standard care and the cost of CF care. The age-specific annual drug costs of standard care treatment were calculated based on a retrospective claims study of US health-care costs and utilisation among patients with CF. 58 Three sources59–61 were used to obtain relationships between costs, age and disease severity category. Three disease severity categories were defined: mild (percentage predicted FEV₁ ≥ 70%), moderate (percentage predicted FEV₁ of 40–69%) and severe (percentage predicted FEV₁ < 40%). The relative proportions in each age group are multiplied by the costs per disease severity category. US dollars were converted to UK pounds.

Cost of ivacaftor

The drug costs of standard care and the CF costs are the same in the ivacaftor–standard care strategy as in the standard care strategy. The annual cost of ivacaftor was set at £182,000 (Vertex 2012) (Table 22). The number of years until the patent of ivacaftor expires was assumed to be 14 years; the manufacturer assumed that after this a generic drug would be launched at a lower price of £20,000 (Vertex 2012). 16 The manufacturer used an adherence rate for ivacaftor of 91% based on the Phase III trials,21,27 but they indicated that real-world adherence rates are typically lower than those observed in clinical studies. The observed efficacy of ivacaftor is based on the observed adherence rate of 91%; a reduction of adherence would be likely to impact on efficacy.

Costs of lung transplantation

For the lung transplantation part of the model, two cost estimates are relevant: the cost of the transplantation itself and the costs of follow-up (Table 23). For the cost of the transplantation we used the 2010 reference costs. 65 We combined the costs for elective in-hospital stay with the costs of excess elective hospital days, which resulted in an estimate of £42,018 per transplantation.

The costs of follow-up were based on a study by Anyanwu et al. 66 In this study, costs are reported for up to 15 years after transplantation (between years 10 and 15 based on extrapolation). We used the costs as reported for bilateral transplantation as these are most common (in 2010, 26 out of 29 transplants) in CF patients. As these costs were reported in 1999 UK pounds and were discounted at 6%, we first reversed the discounting and then adjusted the 1999 UK pounds to 2011 UK pounds using a price index of 1.48. 67,68 It is important to realise that the costs per year reported by Anyanwu et al. 66 are per transplanted patient, whereas our model required input per patient still alive. Thus, the costs reported by Anyanwu et al. 66 were adjusted according to the survival rates in the same study. Note that once patients have had a lung transplant, only the above-described follow-up costs apply and thus costs for treatment of CF were assumed to be zero.

Model assumptions

The main model assumptions made in the cost-effectiveness analysis are summarised below:

• The starting patient population for the individual patient simulation is the population included in the two RCTs included in the clinical effectiveness review. We assumed that this population is representative of the total CF population in England.

• The efficacy of ivacaftor will translate into better survival based on a survival function presented by Liou et al. 37

• We assumed that the population under investigation was comparable with the population from Liou et al. 37

• The age-specific annual decline in percentage predicted FEV₁ was based on the epidemiologic study of CF in a large population of patients with CF in the USA and Canada. We assumed that these declines were also appropriate for the UK population.

• Patients treated with standard care experience a decline in FEV₁ as reported in the RCTs for the first 96 weeks; thereafter an age-dependent annual decline based on a large epidemiologic study was used.

• The weight-for-age z-score was assumed to be constant over a lifetime period from the baseline score.

• The annual exacerbation rate was assumed to be dependent on percentage predicted FEV₁ and age.

• For patients < 12 years old, no reduction in exacerbations was assumed, as exacerbations were rare in both treatment groups.

• The QoL of patients after a lung transplant was based on the patients who have undergone a bilateral transplant as the majority of transplants are bilateral.

• Average dosages are used to estimate the costs of the high-cost drugs.

• The annual cost of ivacaftor was assumed to be £182,000.

• The number of years until the ivacaftor patent expires was assumed to be 14 years.

• The annual cost of a generic drug was assumed to be £20,000.

• The adherence rate to ivacaftor was assumed to be 91% based on the Phase III trials.

Conservative scenario

In the conservative scenario, the percentage predicted FEV₁ of ivacaftor-treated patients stays stable for 96 weeks, after which it declines by the same rate as in the standard care population. Without discounting, treatment with ivacaftor leads to a lifetime additional cost of £2.1M (standard care £0.4M, ivacaftor + standard care £2.5M) while gaining 2.7 life-years (standard care 16.25, ivacaftor + standard care 18.94) or 2.18 QALYs (standard care 12.29, ivacaftor + standard care 14.47). After discounting costs and effects, the additional costs of ivacaftor amount to £1.6M, with a gain in life-years of 1.52 and a gain in QALYs of 1.27, leading to an ICER of £1.27M. The ratio of costs to effects is least favourable for the younger age groups, becoming more favourable as the age at start of treatment increases (Table 27). The PSA suggests that the ICER is likely to be between £980,000 and £1.85M per QALY gained (see Table 30).

Optimistic scenario

In the optimistic scenario, the percentage predicted FEV₁ of ivacaftor-treated patients stays stable over lifetime, while in standard care patients the percentage predicted FEV₁ declines over time. Without discounting, treatment with ivacaftor leads to a lifetime additional cost of £2.5M (standard care £0.4M, ivacaftor + standard care £2.9M) while gaining 19.8 life-years (standard care 16.25, ivacaftor + standard care 36.00) or 16.3 QALYs (standard care 12.29, ivacaftor + standard care 28.66). After discounting costs and effects, the additional costs of ivacaftor amount to £1.8M, with a gain in life-years of 6.20 and a gain in QALYs of 5.26, leading to an ICER of £334,775. The ratio of costs to effects is most favourable for the younger age groups and increases slightly as the age at the start of treatment increases (Table 28). The PSA suggests that, given the parameter uncertainty, the ICER is likely to be between £284,000 and £401,000 per QALY gained (see Table 30).

Intermediate scenario

The intermediate scenario lies between the conservative and optimistic scenarios, with the percentage predicted FEV₁ of ivacaftor-treated patients declining after 96 weeks at a rate of 66% of that of standard care patients. Without discounting, treatment with ivacaftor leads to a lifetime additional cost of £2.2M (standard care £0.4M, ivacaftor + standard care £2.6M) while gaining 5.29 life-years (standard care 16.25, ivacaftor + standard care 21.54) or 4.27 QALYs (standard care 12.29, ivacaftor + standard care 16.56). After discounting costs and effects, the additional costs of ivacaftor amount to £1.7M, with a gain in life-years of 2.58 and a gain in QALYs of 2.16, leading to an ICER of £771,297. As in the conservative scenario, the ratio of costs to effects is least favourable for the younger age groups and becoming more favourable as the age at the start of treatment increases (Table 29). The PSA suggests that the ICER is likely to be between £607,699 and £1.05M per QALY gained (Table 30).

Subgroup analysis: ‘optimistic scenario’ in young children with good lung function

Tables 31 to 34 show that assuming no further disease progression for the proportion of the trial population who were aged < 12 years and had specified baseline lung function would reduce the ICER substantially (range £154,257 to £200,268) in comparison with the optimistic scenario for the whole population, which had an ICER of £334,775 (see Optimistic scenario). This is for three main reasons: reduced mortality (and thus increased life expectancy), increased utility (and thus increased QALYs) and reduced cost of standard care (to zero in scenarios 1 and 3).

Clinical effectiveness

The clinical effectiveness review found that ivacaftor is an effective treatment for adults and children with the G551D mutation based on two RCTs, one in adults and one in children, and an open-label extension trial of the two included RCTs. The studies were generally well conducted and were rated as low or unclear on all risk of bias domains; limited details available on the study in children resulted in some domains being rated as unclear. All outcomes assessed in the review showed greater improvements in the ivacaftor group than in the placebo group. All differences were statistically significant (p < 0.02) with the exception of QoL in children where differences between ivacaftor and placebo favoured ivacaftor but failed to reach statistical significance at either 24- or 48-week follow-up. Ivacaftor was associated with an absolute increase of around 10% in the percentage predicted FEV₁ compared with baseline values while levels in patients treated with placebo stayed around the same. Results from the open-label study showed that improvements in lung function, quality of life and weight were maintained after a further 48 weeks of treatment with ivacaftor (96 weeks’ total treatment). Subgroup analysis showed similar improvements in lung function for all subgroups investigated. This suggests that the effects of ivacaftor do not differ significantly according to age, sex, geographical region or baseline lung function. Ivacaftor does not appear to be associated with an increased risk of adverse events or withdrawals.

Cost-effectiveness

The economic evaluation of ivacaftor showed that the ICER varies between £334,000 and £1.27M per QALY gained, depending on the assumptions made for the long-term effectiveness of ivacaftor. We explored three scenarios: conservative, optimistic and an intermediate scenario. The variation between ICERs was mostly due to large differences in QALYs gained between the scenarios, which varied between 1.27 and 5.26. We found that the impact of the remaining parameter uncertainty is small compared with the uncertainty caused by the long-term extrapolation. An additional optimistic scenario for the subgroup of patients < 12 years of age with no or little lung damage resulted in an ICER of between £154,000 and £200,000 per QALY gained.

We also explored the budget impact for England of introducing ivacaftor to all eligible CF patients. We found that the total additional lifetime costs (discounted) for this cohort would amount to £438M to £479M, whereas the lifetime costs for standard care only would amount to £72M. The total additional costs in the first year would amount to £43M (including the costs for genetic testing) while the costs for standard care would amount to only £5.6M.

When the population treated with ivacaftor is limited to patients < 12 years of age with no or little lung damage, we found that the total additional lifetime costs (discounted) amount to £51M to £113M, whereas the lifetime costs for standard care only would be between £9M and £17M. The total additional costs in the first year would amount to £3.7M to £7.5M (including the costs for genetic testing) while the costs for standard care only would amount to £0.4M to £0.8M.

Clinical effectiveness

Extensive literature searches were conducted in an attempt to maximise retrieval of relevant studies. These included electronic searches of a variety of bibliographic databases, screening references of included publications, as well as screening of clinical trials registers and conference abstracts to identify unpublished studies. Despite this, we were able to identify only two RCTs and one open-label study which met the inclusion criteria for our review. This finding is to be expected because ivacaftor is a very new drug and so has currently been evaluated only by the manufacturer.

Clear inclusion criteria were specified in the protocol for this review. The eligibility of studies for inclusion is therefore transparent. In addition, we have provided specific reasons for excluding all of the studies which were assessed as full-text papers. We restricted inclusion into the review to studies that reported a minimum of 3 months’ follow-up. We felt that follow-up shorter than this was insufficient to establish a sustained treatment effect. This resulted in the exclusion of three small Phase II trials, all of which reported positive effects of ivacaftor. The review process followed recommended methods to minimise the potential for error and/or bias. 18 Search results were independently screened for relevance by two reviewers, and full-text inclusion assessment, data extraction and quality assessment were done by one reviewer and checked by a second. Any disagreements were resolved by consensus. We conducted a formal quality assessment to evaluate the risk of bias in the included studies. This was limited by the lack of published details on the children’s trial, which resulted in this trial being rated as unclear on some domains.

The small number of studies conducted in different patient groups (adults and children) meant that it was not possible to conduct a formal meta-analysis. Instead results were presented grouped on outcome to provide an overview of the evidence available for each outcome. The small number of included studies also meant that it was not possible to formally investigate the potential for publication bias. However, the risk of publication bias is likely to be very low. All trials of ivacaftor identified by the searches (both included and excluded) were conducted by the manufacturers and were registered on trial registries, which were searched as part of our systematic searches. It would not have been possible for anyone other than the manufacturer to conduct a trial of ivacaftor prior to it being licensed. Licensing was only granted in the USA on 31 January 2012,14 and so there would not have been time for new studies to be conducted since these data and for results to have been published that would have fulfilled our inclusion criteria. Details were available from multiple sources for the two included RCTs. The primary report of the adults’ study included a supplementary file containing detailed results information. However, some information was lacking from this report, in particular information on the variability of effect estimates (CIs and standard deviations). Other sources were therefore used to locate this information, in particular documents available via the FDA website34–36 and supplementary data provided by Vertex. The children’s study and the open-label study have not yet been reported as full-text journal articles. Information relating to these studies was therefore obtained from conference abstracts,27 the FDA website,34–36 a Vertex press release,29 and the confidential dossier supplied by the manufacturer. 16 A problem with obtaining results from multiple sources was that there were slight discrepancies in figures reported in different reports. We therefore developed a hierarchy to select a single estimate to contribute to the results of the review. We also included details of differences in results in Appendix 5. These differences were all small and unlikely to have impacted on the conclusions of the review.

The methods of analysis in the included RCTs appeared statistically robust. This was supported by the FDA analysis of these studies, which included various sensitivity analyses using different statistical methods. These analyses all showed similar results for the primary outcome of percentage predicted FEV₁ through to week 24. 36

Cost-effectiveness

We performed a critical appraisal of the model submitted by the manufacturer, and were able to improve many model inputs, such as the quality-of-life estimates, the costs of standard care and the natural decline in percentage predicted FEV₁. We were also able to include lung transplantation (with its associated survival, quality-of-life and follow-up costs) to the model.

The patient population in the model was based on the patients included in the two RCTs. Comparison of baseline characteristics of these patients with characteristics of all CF patients in the UK obtained from the UK CF registry showed that the patient populations were comparable with respect to the proportion of male to female and the median FEV₁. However, the percentage of patients with a chronic P. aeruginosa infection was higher in our patient population (76% in the adult population) compared with 51% in the total UK CF population aged > 12 years. This could indicate that despite the comparable FEV₁ the patient population in the model included more patients whose lungs were already severely and irreversibly damaged by CF, which may have an influence on the generalisability of the results.

The survival function used in the model requires an estimate of the annual number of exacerbations per patient. The annual exacerbation rates for the standard care group used in the manufacturer model, 1.4 for patients ≥ 12 years and 0 for patients < 12 years, were based on the small numbers of patients in the two trials (n = 78 and n = 28, respectively) and an observation period of 48 weeks. We adapted these rates because they were based on small sample sizes and obtained from a trial population, which is generally a selective population due to the inclusion and exclusion criteria used. The current clinical trials, for example, included only patients with clinically stable CF without any respiratory infection or exacerbation in the 4 weeks before the start of the study. Furthermore, studies reporting on exacerbation frequencies should at least have a follow-up of 1 year to account for seasonal variation. Therefore, the annual exacerbation rates in the adapted model were based on a large, continuous, observational registry, the US CF Patient Registry, which most likely resulted in exacerbation rates that were more representative for the total CF population than the trial estimates.

In the model, we have included lung transplantation, assuming that patients with a percentage predicted FEV₁ < 30% are eligible for a lung transplant. However, in reality, eligibility is increasingly based on additional factors as, in the past years, the survival of patients with a percentage predicted FEV₁ < 30% has markedly increased. 71 Additionally, the costs of follow-up of transplanted patients were based on a study by Anyanwu et al. 66 However, it is unclear if these costs also include the potential CF-related costs that will still occur for CF complications in other organs. Thus, it is possible that the post-transplantation costs should be higher. However, both the percentage of patients eligible for lung transplant and the post-transplantation costs have a very minimal impact on the ICER.

Three out of four dimensions on which ivacaftor showed an effect (percentage predicted FEV₁, weight and exacerbations) were taken into account in the model. However, the decrease in the number of exacerbations due to ivacaftor was included in the model only in so far as it affected the survival of the patients. A reduction in exacerbations results in decreased mortality and so reduced exacerbations contributes to life-year gains. It is reasonable to assume that a reduction in exacerbations also has a direct effect on lung function, QoL and costs. Studies by Saunders et al. 72,73 show that there is a strong association between frequency of exacerbations and a decline in lung function. In addition, a study by Britto et al. 74 found that pulmonary exacerbations in the past 6 months were negatively associated with both physical and psychosocial measures of quality of life. Furthermore, treatment of pulmonary exacerbations is associated with high health-care costs, especially for inpatient hospital care and medication. 75 A reduction in exacerbations would therefore lead to an increase in quality of life and a reduction of health-care costs. However, due to a lack of valid data on the association between CF exacerbations and QoL (in terms of utility values) and the costs of a CF exacerbation, the independent impact of exacerbations on QoL and costs could not be included in the model. The CF bandings cover most costs for patients with CF but it was not possible to disentangle maintenance costs from treatment costs for exacerbations. Ideally, we would have added an event probability to the model, indicating per cycle what the probability of an exacerbation is, possibly dependent on age and percentage predicted FEV₁. However, such data were neither found in the literature nor available in the clinical study reports of ivacaftor. One of the challenges in modelling exacerbations explicitly is that a treatment-related reduction in exacerbations is often a result of both an indirect impact through lung function and a direct impact. In order to avoid double counting, we would have needed patient-level data on all lung function measurements and all dates when exacerbations occurred. Also, information about the severity of the exacerbation would be required. Additionally, valid data would need to be found on the costs of a CF exacerbation. In the data source used as input for the cost of CF care by severity no distinction was made between costs for maintenance treatment and costs for exacerbations. If the treatment effects on exacerbations could have been taken into account then the gain in QALYs in the ivacaftor group might have been higher and the savings in CF-related health-care costs might have been higher, resulting in a lower ICER.

In the model, quality-of-life values and costs were assumed to be dependent on disease severity defined in terms of percentage predicted FEV₁. However, a study by Gee et al. 76 showed that this clinical measure explains only part of the variation in QoL and they suggested that other clinical and social factors might also be important, such as social support and coping strategies. Our analysis of the cost data from the CF registry showed the same, that only 26% of variance was explained by age and percentage predicted FEV₁. As always in modelling diseases, further refinements of the health states considered would provide a better reflection of the heterogeneity among patients, but as a result it would likely become more difficult to find the data required to inform transitions between health states.

The costs of standard care included in the model are not complete. The CF tariff bands do not include surgery for certain device implantations nor do they include costs of medications prescribed by the general practitioner. Thus, the true costs of standard care in CF will be higher than the costs used in the model, which would decrease the ICER. However, the impact of these additional costs for standard care would probably have a minimal impact on the ICER given that the annual cost of standard care was only a very small fraction of total annual cost given the price of ivacaftor of £182,000.

One of the main challenges in estimating lifetime cost and QALYs is extrapolating from short-term trial data, in this case the time horizon for the randomised component being only 48 weeks, although up to 96 weeks with some open-label data. On this basis we constructed a set of scenarios based on various assumptions regarding the degree of maintenance of the treatment effect. We make no claim as to which scenario is more likely. However, given expert opinion that ivacaftor might permanently correct the underling biochemical disorder, it does seem likely that young children with little or no lung damage will do much better. Therefore, we conducted a subgroup analysis on children up to age 12 with little or no permanent lung damage, which assumed they would have length and quality of life as if they had been cured of the CF.

The budget impact analysis was performed only for the cohort of CF patients as observed in the year 2011. The impact of new patients becoming eligible for the treatment with ivacaftor after that year was not taken into account. Lifetime costs for ivacaftor for this group of patients would have been lower, because these patients will be treated for fewer years before the drug comes off patent.

Clinical effectiveness

The main uncertainty relates to the long-term clinical effectiveness of ivacaftor. The available RCTs were only 48 weeks in duration. The open-label trial provided information on an additional 48 weeks of ivacaftor treatment in adults and 24 weeks in children, meaning that data are currently available for 96 weeks of treatment with ivacaftor in adults and 72 weeks of treatment in children. The open-label trial is intended to run for 96 weeks; when full data are available from this study, information will be available on the effectiveness of a total of 144 weeks’ (just over 2.5 years’) treatment with ivacaftor in adults and children.

Ivacaftor has been evaluated only in adults and children ≥ 6 years. Its potential effect in children younger than this is unclear. It may be that treating patients at a very young age with ivacaftor is more effective as it may prevent some of the complications of CF from developing. Ivacaftor works by correcting the underlying deficit in chloride ion transport. This is supported by the results of the systematic review which show improvements in sweat chloride levels, with levels in ivacaftor-treated patients returning to within normal ranges (i.e. below the threshold required for a diagnosis of CF) within 2 weeks of treatment with ivacaftor. It would therefore appear reasonable to hypothesise that treating very young patients in whom no lung damage or infection has yet developed may prevent symptoms of CF from ever developing. However, until trials with long-term follow-up are done in this age group of children the potential harms and benefits remain uncertain. The trials to date focus on the impact of ivacaftor on lung function. If ivacaftor works by correcting the CFTR defect and is given before damage has occurred to other organs such as the pancreas, then it may also prevent damage to these organs, thereby preventing the occurrence of complications associated with CF, such as diabetes. Further evidence on the effects of ivacaftor on other organs is required to address this issue.

The trials evaluated in this review were restricted to patients with the G551D mutation. These represent only around 5.7% of the UK population with CF. A study of 140 patients homozygous for the ∆F508 mutation did not find a significant difference in percentage predicted FEV₁ between patients treated with ivacaftor and those receiving placebo after 16 weeks of treatment. Adverse events were similar between the groups and there were some small benefits of ivacaftor on other outcomes. 77,78 A further study is ongoing which is investigating ivacaftor in combination with VX-809, an investigational CFTR corrector, in patients with CF and homozygous for the ∆F508 mutation. 79 This trial is still ongoing but early results suggest potentially beneficial effect of the drug combination. If this combination is proved to be effective it would considerably expand the potential usage of ivacaftor as the ∆F508 is the most common CF-causing mutation in the UK population. 7

Cost-effectiveness

From a cost-effectiveness perspective the long-term effectiveness is also the main uncertainty. The various scenarios explored for this long-term effectiveness show a wide range of ICERs. Only when longer-term data on ivacaftor become available will it be clear which of these ICERs is most relevant.

Additional uncertainty is caused by the long-term costs of ivacaftor. In the model, it is assumed that, after 14 years, a generic version of ivacaftor will be available at £20,000 (as opposed to £182,000 for the branded version). This information was obtained from the manufacturer and therefore assumed to be best possible estimates. It is clear that the costs of ivacaftor are the main driver of the results, and thus the lifelong cost estimates should be interpreted with care. If the patent expiry for ivacaftor were to occur sooner or if the generic costs are lower, the lifetime costs for treatment with ivacaftor would be lower, resulting in a lower cost-effectiveness ratio compared with standard care.

Adoption of a societal perspective might have been interesting; however, this was beyond the scope of the review, which was to estimate the cost-effectiveness of ivacaftor from the perspective of the NHS. However, had such an analysis been undertaken it is likely that savings in costs for informal care and productivity loss in those treated with ivacaftor would have reduced the ICER.

Also, given that ivacaftor is an orphan drug, there is no clear benchmark to indicate whether or not ivacaftor should be considered cost-effective. Several other orphan drugs are in current use despite ICERs being considerably higher than the threshold of £20,000 to £30,000 applied in most NICE appraisals;69 some examples of these are shown in Table 35.

EMBASE (OvidSP) 1974–2012 week 17

Searched 3 May 2012.

1. Ivacaftor/ (72)

2. (Ivacaftor or Kalydeco or VX-770 or VX770 or 873054-44-5 or ivacaftorum).af. (138)

3. or/1-2 (138)

MEDLINE (OvidSP) 1946–April 2012 week 4

Searched 3 May 2012.

1. (Ivacaftor or Kalydeco or VX-770 or VX770 or 873054-44-5 or ivacaftorum).af. (10)

MEDLINE In-Process & Other Non-Indexed Citations (OvidSP) up to 2 May 2012

MEDLINE Daily Update (OvidSP) up to 2 May 2012

Searched 3 May 2012.

1. (Ivacaftor or Kalydeco or VX-770 or VX770 or 873054-44-5 or ivacaftorum).af. (3)

Cochrane Database of Systematic Reviews (CDSR) (Wiley Online Library) up to 2012 issue 4

Cochrane Central Register of Controlled Trials (CENTRAL) (Wiley Online Library) up to 2012 issue 4

Database of Abstracts of Reviews of Effects (DARE) (Wiley Online Library) up to 2012 issue 2

Health Technology Assessment (HTA) Database (Wiley Online Library) up to 2012 issue 2

NHS Economic Evaluation Database (NHS EED) (Wiley Online Library) up to 2012 issue 2

Searched 3 May 2012.

#1 (Ivacaftor OR Kalydeco OR VX-770 OR VX770 OR VX 770 OR 873054-44-5 OR ivacaftorum)

CDSR retrieved no records.

CENTRAL retrieved five records.

DARE retrieved no records.

HTA retrieved no records.

NHS EED retrieved 0 records.

Database of Abstracts of Reviews of Effects (DARE) (CRD) up to 3 May 2012

Health Technology Assessment (HTA) Database (CRD) up to 3 May 2012

NHS Economic Evaluation Database (NHS EED) (CRD) up to 3 May 2012

Searched 3 May 2012.

#1 (Ivacaftor OR Kalydeco OR VX-770 OR VX770 OR VX 770 OR 873054-44-5 OR ivacaftorum)

DARE retrieved no records.

HTA retrieved no records.

NHS EED retrieved no records.

Latin American and Caribbean Health Sciences Literature (LILACS) (VHL)

URL: http://lilacs.bvsalud.org/en

Searched 4 May 2012.

Ivacaftor OR Kalydeco OR VX-770 OR VX770 OR 873054-44-5 OR ivacaftorum (0)

ClinicalTrials.gov

URL: http://clinicaltrials.gov/ct2/search/advanced

Searched 4 May 2012.

Ivacaftor OR Kalydeco OR VX-770 OR VX770 OR VX 770 OR 873054-44-5 OR ivacaftorum (18)

mRCT – metaRegister of Controlled Trials (internet)

URL: www.controlled-trials.com/mrct/search.html

Searched 4 May 2012.

Ivacaftor OR Kalydeco OR VX-770 OR VX770 OR VX 770 OR 873054-44-5 OR ivacaftorum (12)

World Health Organization International Clinical Trials Registry Platform (ICTRP) (internet)

URL: www.who.int/ictrp/en

Searched 4 May 2012.

Ivacaftor OR Kalydeco OR VX-770 OR VX770 OR VX 770 OR 873054-44-5 OR ivacaftorum (18)

European Cystic Fibrosis Society (ECFS) Conferences – searched titles (no abstracts available)

URL: www.ecfs.eu/meetings/ecfs

NACFC: The Annual North American Cystic Fibrosis Conferences

URL: www.nacfconference.org

CIPP: International Congress on Pediatric Pulmonology

URL: www.cipp-meeting.org/index.htm

Searched titles (no abstracts available).

Update searches

MEDLINE (OvidSP) 2002–May 2012 week 1

Searched 11 May 2012.

1. cystic fibrosis/ (26,164)

2. ((Cystic adj2 fibrosis) or mucoviscidosis).ti,ab,ot,hw. (34,176)

3. CF.ti,ot. (1276)

4. (pancreas adj3 fibrocystic adj3 disease$).ti,ab,ot,hw. (16)

5. or/1-4 (34,859)

6. economics/ (26,269)

7. exp "costs and cost analysis"/ (164,214)

8. economics, dental/ (1840)

9. exp "economics, hospital"/ (17,877)

10. economics, medical/ (8463)

11. economics, nursing/ (3861)

12. economics, pharmaceutical/ (2325)

13. (economic$ or cost or costs or costly or costing or price or prices or pricing or pharmacoeconomic$).ti,ab. (359,328)

14. (expenditure$ not energy).ti,ab. (15,019)

15. (value adj1 money).ti,ab. (18)

16. budget$.ti,ab. (15,258)

17. or/6-16 (475,664)

18. ((energy or oxygen) adj cost).ti,ab. (2417)

19. (metabolic adj cost).ti,ab. (636)

20. ((energy or oxygen) adj expenditure).ti,ab. (13,970)

21. or/18-20 (16,385)

22. 17 not 21 (471,962)

23. letter.pt. (745,733)

24. editorial.pt. (297,634)

25. historical article.pt. (282,385)

26. or/23-25 (1,312,332)

27. 22 not 26 (446,294)

28. 5 and 27 (667)

29. animals/ not (animals/ and humans/) (3,620,832)

30. 28 not 29 (657)

31. remove duplicates from 30 (639)

32. limit 31 to yr="2002 -Current" (307)

MEDLINE In-Process & Other Non-Indexed Citations (OvidSP) 2002–10 May 2012

MEDLINE Daily Update (OvidSP) 2002–10 May 2012

Searched 11 May 2012.

1. cystic fibrosis/ (13)

2. ((Cystic adj2 fibrosis) or mucoviscidosis).ti,ab,ot,hw. (860)

3. CF.ti,ot. (153)

4. (pancreas adj3 fibrocystic adj3 disease$).ti,ab,ot,hw. (2)

5. or/1-4 (999)

6. economics/ (3)

7. exp "costs and cost analysis"/ (127)

8. economics, dental/ (0)

9. exp "economics, hospital"/ (16)

10. economics, medical/ (0)

11. economics, nursing/ (0)

12. economics, pharmaceutical/ (2)

13. (economic$ or cost or costs or costly or costing or price or prices or pricing or pharmacoeconomic$).ti,ab. (28,438)

14. (expenditure$ not energy).ti,ab. (752)

15. (value adj1 money).ti,ab. (2)

16. budget$.ti,ab. (1477)

17. or/6-16 (29,977)

18. ((energy or oxygen) adj cost).ti,ab. (158)

19. (metabolic adj cost).ti,ab. (44)

20. ((energy or oxygen) adj expenditure).ti,ab. (649)

21. or/18-20 (833)

22. 17 not 21 (29,739)

23. letter.pt. (18,035)

24. editorial.pt. (11,136)

25. historical article.pt. (158)

26. or/23-25 (29,322)

27. 22 not 26 (29,403)

28. 5 and 27 (27)

29. animals/ not (animals/ and humans/) (2139)

30. 28 not 29 (27)

31. remove duplicates from 30 (27)

32. limit 31 to yr="2002 -Current" (22)

EMBASE (OvidSP) 2002–2012 week 18

Searched 11 May 2012.

1. cystic fibrosis/ (41,722)

2. ((Cystic adj2 fibrosis) or mucoviscidosis).mp. (48,828)

3. CF.ti,ot. (2503)

4. (pancreas adj3 fibrocystic adj3 disease$).mp. (16)

5. or/1-4 (49,844)

6. health-economics/ (31,568)

7. exp economic-evaluation/ (183,852)

8. exp health-care-cost/ (176,126)

9. exp pharmacoeconomics/ (152,648)

10. or/6-9 (423,153)

11. (econom$ or cost or costs or costly or costing or price or prices or pricing or pharmacoeconomic$).ti,ab. (504,978)

12. (expenditure$ not energy).ti,ab. (20,285)

13. (value adj2 money).ti,ab. (1089)

14. budget$.ti,ab. (20,942)

15. or/11-14 (526,198)

16. 10 or 15 (774,146)

17. letter.pt. (784,623)

18. editorial.pt. (406,289)

19. note.pt. (515,518)

20. or/17-19 (1,706,430)

21. 16 not 20 (696,305)

22. (metabolic adj cost).ti,ab. (744)

23. ((energy or oxygen) adj cost).ti,ab. (2877)

24. ((energy or oxygen) adj expenditure).ti,ab. (17,302)

25. or/22-24 (20,179)

26. 21 not 25 (691,788)

27. exp animal/ (1,778,262)

28. exp animal-experiment/ (1,613,173)

29. nonhuman/ (3,832,807)

30. (rat or rats or mouse or mice or hamster or hamsters or animal or animals or dog or dogs or cat or cats or bovine or sheep).ti,ab,sh. (4,657,237)

31. or/27-30 (6,588,417)

32. exp human/ (13,530,525)

33. exp human-experiment/ (300,208)

34. 32 or 33 (13,531,960)

35. 31 not (31 and 34) (52,04,957)

36. 26 not 35 (643,011)

37. 5 and 36 (1330)

38. limit 37 to embase (1088)

39. remove duplicates from 38 (1086)

40. limit 39 to yr="2002 -Current" (745)

NHS Economic Evaluation Database Centre for Reviews and Dissemination 2002–3 May 2012

Searched 3 May 2012.

1. MeSH DESCRIPTOR Cystic Fibrosis (86)

2. (Cystic near2 fibrosis) (271)

3. (mucoviscidosis) (0)

4. (pancreas near3 fibrocystic near3 disease*) (0)

5. #1 OR #2 OR #3 OR #4 (271)

6. (#5) IN NHSEED (52)

7. (#6) FROM 2002 TO 2012 (22)

Health Economic Evaluations Database (HEED) (Wiley Online Library) 2002–21 May 2012

Searched 21 May 2012.

Using compound search.

1. All Data: ’CYSTIC FIBROSIS’ OR ’mucoviscidosis’ OR ’CF’ AND

2. Journal Date: >= 2002

Sixty-five references were retrieved.

MEDLINE (OvidSP) 1946–May 2012 week 1

Searched 11 May 2012.

1. cystic fibrosis/ (26,164)

2. ((Cystic adj2 fibrosis) or mucoviscidosis).ti,ab,ot,hw. (34,176)

3. CF.ti,ot. (1276)

4. (pancreas adj3 fibrocystic adj3 disease$).ti,ab,ot,hw. (16)

5. or/1-4 (34,859)

6. (sf36 or sf 36 or short form 36 or shortform 36).ti,ab. (12,308)

7. (sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirty six or short form thirtysix or short form thirty six).ti,ab. (1)

8. (sf6 or sf 6 or short form 6 or shortform 6 or sf six or sfsix or shortform six or short form six).ti,ab. (890)

9. (euroqol or euro qol or eq5d or eq 5d).ti,ab. (2653)

10. (hql or hrql or hqol or h qol or hrqol or hr qol).ti,ab. (7311)

11. (hye or hyes).ti,ab. (52)

12. health$ year$ equivalent$.ti,ab. (36)

13. (hui or hui1 or hui2 or hui3 or hui4 or hui-4 or hui-1 or hui-2 or hui-3).ti,ab. (692)

14. (quality of well being or quality of wellbeing or qwb).ti,ab. (315)

15. (Disability adjusted life year$ or Disability-adjusted life year$ or health adjusted life year$ or health-adjusted life year$ or years of healthy life or healthy years equivalent or years of potential life lost or years of health life lost or quality adjusted life year$).ti,ab. (5478)

16. (QALY$ or HRQOL or HRQL or DALY$ or HALY$ or YHL or HYES or YPLL or YHLL).ti,ab. (11,747)

17. (health$ adj3 utilit$).ti,ab. (1526)

18. (Time trade-off or time tradeoff or TTO or Standard gamble).ti,ab. (1266)

19. ((Cystic Fibrosis adj2 Questionnaire$) or CFQ).ti,ab. (146)

20. or/6-19 (28,430)

21. 5 and 20 (127)

22. animals/ not (animals/ and humans/) (3,620,832)

23. 21 not 22 (127)

24. remove duplicates from 23 (125)

MEDLINE In-Process & Other Non-Indexed Citations (OvidSP) up to 10 May 2012

EMBASE (OvidSP) 1974–2012 week 18

Searched 11 May 2012.

1. cystic fibrosis/ (41,722)

2. ((Cystic adj2 fibrosis) or mucoviscidosis).mp. (48,828)

3. (pancreas adj3 fibrocystic adj3 disease$).mp. (16)

4. CF.ti,ot. (2503)

5. or/1-4 (49,846)

6. (sf36 or sf 36 or short form 36 or shortform 36).ti,ab. (17,443)

7. (sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirty six or short form thirtysix or short form thirty six).ti,ab. (1)

8. (sf6 or sf 6 or short form 6 or shortform 6 or sf six or sfsix or shortform six or short form six).ti,ab. (1381)

9. (euroqol or euro qol or eq5d or eq 5d).ti,ab. (4305)

10. (hql or hrql or hqol or h qol or hrqol or hr qol).ti,ab. (10,640)

11. (hye or hyes).ti,ab. (62)

12. health$ year$ equivalent$.ti,ab. (41)

13. (hui or hui1 or hui2 or hui3 or hui4 or hui-4 or hui-1 or hui-2 or hui-3).ti,ab. (935)

14. (quality of well being or quality of wellbeing or qwb).ti,ab. (365)

15. (Disability adjusted life year$ or Disability-adjusted life year$ or health adjusted life year$ or health-adjusted life year$ or years of healthy life or healthy years equivalent or years of potential life lost or years of health life lost or quality adjusted life year$).ti,ab. (7550)

16. (QALY$ or HRQOL or HRQL or DALY$ or HALY$ or YHL or HYES or YPLL or YHLL).ti,ab. (17,567)

17. (health$ adj3 utilit$).ti,ab. (2194)

18. (Time trade-off or time tradeoff or TTO or Standard gamble).ti,ab. (1639)

19. ((Cystic Fibrosis adj2 Questionnaire$) or CFQ).ti,ab. (272)

20. or/6-19 (41,148)

21. 5 and 20 (251)

22. animal/ or animal experiment/ (3,374,417)

23. (rat or rats or mouse or mice or murine or rodent or rodents or hamster or hamsters or pig or pigs or porcine or rabbit or rabbits or animal or animals or dogs or dog or cats or cow or bovine or sheep or ovine or monkey or monkeys).mp. (5,438,846)

24. or/22-23 (5,438,846)

25. exp human/ or human experiment/ (13,531,960)

26. 24 not (24 and 25) (4,385,166)

27. 21 not 26 (251)

28. limit 27 to embase (235)

29. remove duplicates from 28 (235)

Cost-effectiveness Analysis Registry (internet)

URL: https://research.tufts-nemc.org/cear4

Searched 14 May 2012.

Searched for ‘Articles’ using the term ‘cystic’ (6)

Searched for ‘Ratios’ using the term ‘cystic’ (14)

Searched for ‘Weights’ using the term ‘cystic’ (23)

A total of 43 references were retrieved.

National Institute for Health and Care Excellence Guidance (internet)

URL: http://guidance.nice.org.uk

Searched 11 May 2012.

Searched for ‘cystic fibrosis’.

Limited to information type: guidance.

Six references retrieved.

TRIP database (internet)

URL: www.tripdatabase.com

Searched 11 May 2012.

Searched for ‘cystic fibrosis’ in title.

Limited to guidelines.

Sixteen references retrieved.

Guidelines International Network (internet)

Searched 11 May 2012

Searched for ‘cystic fibrosis’.

Seven references retrieved.

National Guidelines Clearinghouse (internet)

URL: www.guidelines.gov

Searched 11 May 2012.

Searched for ‘cystic fibrosis’ in title.

Three references retrieved.

US Food and Drug Administration

URL: www.fda.gov

Searched 11 May 2012.

Searched for ‘cystic fibrosis’ in title.

Limited to guidance.

Two references retrieved.

Cystic Fibrosis Trust

URL: www.cftrust.org.uk

Searched 11 May 2012.

Browsed website publications.

Seventeen references retrieved.

Plain English Summary

Cystic Fibrosis (CF) is one of the most common, inherited diseases in white populations. Around 1 in every 2500 babies born in the UK has CF and there are over 9000 people in the UK with CF. CF is caused by a single faulty gene which controls the movement of salt and water in and out of cells. This results in thick sticky mucous clogging up the internal organs (e.g. lungs, pancreas, liver, intestine and reproductive tract) making it difficult to breathe and digest food. Other symptoms can include a troublesome cough, prolonged diarrhoea and poor weight gain. Most of the illness caused by CF is from diseases of the lungs and repeated infections. There is no cure for CF and most treatments (e.g. physiotherapy, antibiotics for infections, drugs to suppress inflammation) target the symptoms rather than the cause of disease. Median survival of the current UK cohort with CF is estimated as 41 years. Most patients die from lung disease. Life expectancy is increasing and is expected to increase to at least 50 years for children born in 2000.

A large number of different mutations have been identified in the gene that causes CF. New treatments are being developed which target specific mutations. Ivacaftor (brand name Kalydeco, Vertex Pharmaceuticals) is the first of these drugs and targets patients with the “G551D” mutation. Around 4.4% of patients with CF in the UK will have at least one G551D mutation. Ivacaftor represents a new approach to treating patients with CF as it targets the underlying cause of CF. It aims to increase salt movement through the cell by targeting a specific protein. Ivacaftor is classed as an “orphan drug” which means that has been developed specifically to treat a rare disease. It has been approved by the American Food and Drug Administration for the treatment of patients with CF who are at least 6 years old and have the G551D mutation. There are currently no similar drugs which target the underlying protein defect in CF on the market.

This review aims to evaluate the effectiveness of ivacaftor tablets for the treatment of cystic fibrosis (CF) in patients age 6 years and older who have at least one G551D mutation. The review will consider both clinical effectiveness (improvement in patients’ symptoms and adverse events) and cost effectiveness (cost of treatment).

2. Decision problem

3. Report methods for synthesis of evidence of clinical effectiveness

We will conduct a systematic review of the evidence on the clinical effectiveness of ivacaftor 150 mg tablet for oral administration for the treatment of cystic fibrosis (CF) in patients age 6 years and older who have at least one G551D mutation in the CFTR gene. The review will follow the general principles recommended in the PRISMA statement and CRD report 4. 17,18

4. Report methods for synthesising evidence of cost effectiveness

5. Timetable/milestones

6. Team members’ contributions

Penny Whiting will be the main reviewer on this project and will maintain day-to-day running of the review. Marie Westwood will act as second reviewer. Both reviewers have contributed to the study protocol and will carry out the study selection, data extraction, analysis and production of the final report. Maiwenn Al will be health economic lead for this project, and thus be responsible for the cost-effectiveness study.

Appendix: Draft search strategy

EMBASE (OvidSP): 1974-2012/wk17

Searched 3.5.12

1. Ivacaftor/ (72)

2. (Ivacaftor or Kalydeco or VX-770 or VX770 or 873054-44-5 or ivacaftorum).af. (138)

3. or/1-2 (138)