Background
Gefitinib is a selective small molecule inhibitor of the epidermal growth factor receptor (EGFR) tyrosine kinase (TK); it is an effective treatment for patients with advanced non small cell lung cancer (NSCLC, stage IIIb/IV, new TNM classification stage IV [
1]) and activating mutations of the EGFR TK [
2‐
5].
The European Medicines Agency (EMA) approval for gefitinib treatment in advanced NSCLC in patients with EGFR mutation-positive (M+) tumours was based largely on evidence from the Iressa Pan-Asia Study (IPASS) [
5], together with a comprehensive review of gefitinib data in EGFR M+ NSCLC patients across lines of therapy. In IPASS, a combination of paclitaxel and carboplatin (Pac/Carb) was compared to gefitinib for first-line treatment of clinically-selected advanced NSCLC patients [
5]. The pre-planned subgroup analysis of the EGFR M+ patients in this study demonstrated that gefitinib had a significantly longer progression free survival (PFS) period than Pac/Carb (HR 0.48 (95% Confidence Interval: 0.36; 0.64), median PFS 9.5 months and 6.3 months, respectively [
5,
6]). Gefitinib was also associated with a lower rate of common terminology criteria (CTC) for grade 3 and 4 adverse events (AE)[
5]. The PFS results and lower incidence of AE for gefitinib were confirmed by other trials in EGFR M+ patients, which compared either Pac/Carb [
3] or other standard doublet chemotherapies such as gemcitabine/cisplatin (Gem/Cis) [
7] or cisplatin/docetaxel [
4] to gefitinib. The increase in median PFS with gefitinib ranged from 1.8 [
7] to 4.9 months [
3].
Though the significant benefit in PFS was clear, there did not appear to be a similar overall survival (OS) benefit of gefitinib over doublet chemotherapy [
3,
4,
7,
8]. In the IPASS, the median OS was 21.6 months for gefitinib and 21.9 for Pac/Carb (p = 0.990)[
8]. One reason for the similar OS could be that all studies allowed for further treatments at disease progression, including a cross-over where patients on chemotherapy could cross-over to gefitinib or another tyrosine kinase inhibitor (TKi) and vice versa [
3‐
5,
7,
8]. Second line therapy will affect OS; this makes it difficult to interpret OS differences between initial treatments. Therefore, in this situation, PFS may be considered a more appropriate measure of the true effect of first-line treatment.
When considering treatment effect from a patient perspective, not only is the length of (progression free) survival important; health-related quality-of-life (HRQoL) during that period is also important. HRQoL was measured with disease-specific HRQoL instruments in two studies; IPASS used the FACT-L and First-SIGNAL used the EORTC QLQ-C30 and QLQ-LC13. Both studies demonstrated an improved HRQoL with gefitinib treatment over doublet chemotherapies [
5,
7,
9].
It is important for a new drug to show added value in comparison to standard care. QALYs are a recognised and established measure of disease burden, including both quantity-of-life (mean life-years) and quality-of-life, and are therefore a useful means of expressing the value of a new therapy. One measure of quality of life is through the evaluation of utilities. Utility is a measure of preference, and ranges from 0 (death) to 1 (full health).
How, then, should gefitinib be evaluated in comparison to standard care? Standard first-line care for advanced NSCLC in the Netherlands is Gem/Cis or pemetrexed/cisplatin (Pem/Cis) doublet chemotherapy [
10]. The CEGEDIM 2008 also reported first-line off label prescription of TKi’s for advanced NSCLC [
10]. While there is a lack of utility data for both gefitinib and standard first-line doublet chemotherapies for advanced NSCLC, the FACT-L data from the IPASS study can be transformed into utility data using a published and widely recognised algorithm [
11]. Furthermore, a recent Dutch study by Grutters et al. evaluated the utilities among survivors of NSCLC [
12] (predominantly stage I, II and IIIa) and found that HRQoL in NSCLC patients is influenced by the occurrence of adverse events and objective response.
The objective of this study is to evaluate the Quality Adjusted PFS of gefitinib versus relevant doublet chemotherapies in the Netherlands in patients with EGFR M+ stage IIIb/IV NSCLC during the progression free state.
Methods
When demonstrating the added value of a treatment, a relevant time period should be used that covers all costs and benefits for that disease. In oncology, a life time horizon is often applied. However, in the NSCLC studies used for this analysis, OS may be largely influenced by the effect of subsequent treatment lines introduced at disease progression. Therefore, this analysis measures the true effect of gefitinib as a first line therapy by exploring preferences/utilities in a Dutch treatment context during the progression free time frame.
Calculation of mean PFS for first-line therapy
In order to calculate the mean PFS for first-line therapy, this analysis uses data from the gefitinib single technology appraisal (STA) submission to the National Institute for Health and Clinical Excellence (NICE) in the UK. For this STA, a cost-effectiveness model was developed to compare gefitinib to other doublet chemotherapies in first-line treatment of EGFR M+ advanced NSCLC patients [
6]. To support this submission, in the absence of a head-to-head trial, a network meta-analysis (NMA) for all standard doublet chemotherapies for stage IIIb/IV NSCLC relative to Pac/Carb [
6] was performed to establish the relative efficacy and safety of treatments. This NMA was based on a systematic literature search performed in May 2009. In the same STA submission, a meta-analysis was performed to estimate the relative effects of gefitinib to Pac/Carb in EGFR M+ patients, using data from the IPASS study [
5] and the North East Japan Study group trial [
3,
6]. The studies reported by Mitsudomi et al. [
4] and Lee et al. [
7] were not used, since they did not use Pac/Carb, but other doublet chemotherapies. The NMA assumed that the relative effect of chemotherapies is not influenced by EGFR mutation status. Table
1 summarises the HRs for PFS and odds ratios for objective response derived in the UK NMA for all treatments of interest [
6].
Table 1
HR, and odds ratios obtained with the NMA for duration of PFS and OR, all relative to Pac/Carb treatment
PFS HR (95% CrI) | 1 | 0.92 (0.80; 1.04) | 0.88 (0.74; 1.05) | 0.43 (0.34; 0.53) |
OR (Odds ratio) (95% CrI) | 1 | 1.16 (0.93; 1.44) | 1.64 (1.15; 2.27) | 4.63 (3.01; 6.98) |
For an economic evaluation, the median PFS was translated into a mean PFS. In order to calculate the mean PFS for all treatments of interest, the HR from Table
1 was applied to an estimated mean PFS for the baseline therapy (Pac/Carb). The mean PFS for Pac/Carb was obtained by extrapolating the median PFS as reported in the IPASS study using a Weibull regression model. In the Technical Support document 14 of DSU about survival analysis [
13], the Weibull distribution is the most commonly used distribution within submissions to NICE. The modelled Weibull curve showed a good fit with the Kaplan Meier PFS curve (almost complete overlap).
Calculation of utility data for the Netherlands
When estimating the preference for a certain health state in the Netherlands, preferences provided by the general Dutch public are needed. Hence, to assess the utilities for Dutch advanced NSCLC EGFR M+ patients, 11 items of the FACT-L questionnaire data from the subgroup of EGFR M+ patients (n = 261) in the progression free period in the IPASS study were converted into Dutch utilities by applying the unequal distribution algorithm published by Lamers et al. [
11].
FACT-L data for each EGFR M+ patient from the IPASS study were obtained for both arms at randomisation, at 1 week of treatment, and then at 3, 6, 9, 12, 15, 18, 24, 30, 36 and 42 weeks, up until progression of disease. For each patient, both the utility at each time point and the change from baseline (CFB) were obtained. The mean difference and standard deviation were calculated and time points were weighted by the number of participants available at each time point. To calculate statistical significance between utility CFB, an unpaired t-test was used.
In the absence of data for all comparators, the utility value calculated for Pac/Carb in the progression free period in IPASS was also used for the other doublet chemotherapies (Pem/Cis and Gem/Cis). This might provide an underestimation of utility values for the other doublet chemotherapies; the implications of this assumption are explored in the discussion.
Calculation of quality adjusted life years in PFS
To estimate the QALYs for the progression free period, the treatment arm-specific utilities were multiplied with the estimated mean PFS to calculate the Quality Adjusted PFS for each treatment.
Since no other utility data for the Netherlands exists we have compared our findings with the utilities used in the model presented in the NICE STA submission [
6]. These utilities are based on British preference weights. The estimation of the utility during PFS from the NICE STA consists of four components: stable baseline disease (i.e. state of disease without any change in pre-treatment condition), objective response, effect of drug administration, and effect of AE. Patients with stable disease have a utility of 0.653 [
14]; objective response gives an increment of 0.053 (mean of Nafees et al. [
14] and Doyle et al. [
15]). The mode of administration of the drug also influenced utility, with a decrement of 0.043 for intravenous therapy and 0.014 for oral therapy [
16].
Adverse events lead to a decrement ranging from 0.03 for rash to 0.09 for neutropenia [
14]; all adverse events were set to occur in the first cycle.
Competing interests
For the last 12 months Dr. B. Biesma declared no conflict of interest. Dr. AM. Dingemans received research funding from AstraZeneca and Roche, she served on the advisory boards of AstraZeneca, Boehringer Ingelheim, Eli Lilly, Roche, and Merck.
Dr. F. Schramel served on the advisory boards of AstraZeneca, and Eli Lilly.
F. van der Scheer and M. Langenfeld are employed by AstraZeneca Netherlands, R. de Peuter and S. Verduyn are employed by Mapi Consultancy, a consultancy which offers services to the pharmaceutical industry.
Authors’ contributions
All authors were in involved in the conception and design of the manuscript, next to this SCV and FvdS were responsible for the acquisition, analysis and interpretation of the data, RdP and ML were involved in interpretation of the data. SCV drafted the manuscript, all other authors made substantial revisions to the manuscript and all authors have given final approval of the version to be submitted. All authors read and approved the final manuscript.