1. Gefitinib at any dose compared with placebo or best supportive care for NSCLC

• General population (Comparison 1)

Three phase III studies (Gaafar 2011 EORTC08021; Kelly 2008 SWOG S0023; Thatcher 2005 ISEL) and a single phase II study (Goss 2009 INSTEP) compared gefitinib with placebo. The ISEL (Thatcher 2005 ISEL), INSTEP (Goss 2009 INSTEP), EORTC 08021 (Gaafar 2011 EORTC08021) and SWOGS0023 (Kelly 2008 SWOG S0023) trials examined survival outcomes, objective response rates and toxicity in the general population. The INSTEP study randomised chemotherapy‐naive patients to 250 mg of gefitinib or placebo as first‐line therapy. The ISEL study studied its effects as second‐line therapy in advanced NSCLC. Detailed subgroup analysis was conducted in the ISEL population and subsequently published. These two studies are also presented below as subgroup analyses (Chang 2006; Hirsch 2006). Subgroups were assessed for evidence by subgroup interactions, thus ensuring that outcomes were indeed different. Pre‐planned subgroup analysis of patients of Asian ethnicity was presented in Chang 2006 and analysis of molecular predictors of outcome was presented in Hirsch 2006. The SWOGS0023 and EORTC08021 studies assessed the effect of gefitinib versus placebo as maintenance therapy after initial treatment. In the SWOG study, patients were included after receiving concurrent cisplatin/etoposide chemotherapy with thoracic radiation (45 Gy, 1.8 Gy per fraction). The EORTC08021 trial included patients not progressing after first‐line platinum doublet chemotherapy. We studied a total of 2605 patients in this group.

• Asian population (Comparison 2)

The INFORM study assessed the use of gefitinib as maintenance therapy in an East Asian patient group (Zhang 2012 INFORM). These patients had achieved disease control after first‐line platinum‐based chemotherapy. Chang 2006 selected only ISEL patients who were of Asian ethnicity. This subgroup represented 20% of the original ISEL population, a total of 342 patients. We included a total of 638 patients in this group.

• EGFR mutation positive population (Comparison 3)

Zhang 2012 INFORM performed planned subgroup analysis on EGFR mutation positive patients and 30 of 79 (38%) tissue tumour samples were positive for EGFR mutations. Hirsch 2006 analysed ISEL tumour biopsy samples to examine the relationships between biomarkers and clinical outcome after gefitinib administration. Two‐hundred and fifteen of 1692 patients (12.7%) in the ISEL trial were assessable for mutation detection. Of these, 26 (12.1%) patients were positive for EGFR mutations. Other biomarkers examined included EGFR gene copy number, EGFR and p‐Akt protein expression and KRAS and BRAF mutations. Data from these other biomarkers are beyond the scope of this review.

2. Gefitinib at any dose compared with other chemotherapeutic agents

We included 18 primary studies in this analysis (Ahn 2012; Crino 2008 INVITE; Cufer 2006 SIGN; Dai 2013; Han 2012 First SIGNAL; Kim 2008 INTEREST; Kim 2016; Lee 2010 ISTANA; Li 2010; Lou 2014; Maemondo 2010 NEJ002; Maruyama 2008 V‐15‐32; Mitsudomi 2010 WJTOG3405; Mok 2009 IPASS; Morere 2010 IFCT‐0301; Sun 2012 KCSG‐LU08‐01; Xu 2015; Yang 2014). Nine of these studies were multicentre, randomised, phase III trials.

These 18 primary studies randomised a total of 5400 patients.

• General population (Comparison 4)

Four studies, SIGN (Cufer 2006 SIGN), INTEREST (Kim 2008 INTEREST), INVITE (Crino 2008 INVITE) and IFCT‐0301 (Morere 2010 IFCT‐0301), compared gefitinib with chemotherapy in 1888 patients and data from these are presented in Comparison 4. Two studies compared gefitinib with first‐line chemotherapy (Crino 2008 INVITE; Morere 2010 IFCT‐0301) and the other two studies compared it with second‐line chemotherapy (Cufer 2006 SIGN; Kim 2008 INTEREST). 'Iressa in NSCLC versus Vinorelbine Investigation in the Elderly' (INVITE) was a randomised, multicentre, phase II trial that compared gefitinib with vinorelbine as first‐line therapy in elderly patients (Crino 2008 INVITE). IFCT‐0301 compared gefitinib, gemcitabine and docetaxel in chemotherapy‐naive patients with a performance status of 2 or 3 (Morere 2010 IFCT‐0301). SIGN (Second‐line Indication of Gefitinib in NSCLC) was a phase II, randomised study comparing gefitinib with docetaxel as second‐line therapy (Cufer 2006 SIGN). INTEREST (Iressa NSCLC Trial Evaluating Response and Survival again Taxotere) was a phase III trial, which assessed the non‐inferiority of gefitinib to docetaxel as second‐line therapy (Kim 2008 INTEREST). Douillard 2010 performed a preplanned secondary analysis to investigate the relationship between biomarkers and clinical outcomes in the INTEREST population. We included a total of 1888 patients in this group.

• Asian population (Comparison 5)

Fourteen studies selected Asian patients only (Ahn 2012; Dai 2013; Han 2012 First SIGNAL; Kim 2016; Lee 2010 ISTANA; Li 2010; Lou 2014; Maruyama 2008 V‐15‐32; Mok 2009 IPASS; Mitsudomi 2010 WJTOG3405; Maemondo 2010 NEJ002; Sun 2012 KCSG‐LU08‐01; Xu 2015; Yang 2014), of which all except six (Ahn 2012; Dai 2013; Kim 2016; Li 2010; Lou 2014; Xu 2015) were phase III studies. We included a total of 3512 patients in this group.

First‐line studies

Five phase III studies (Han 2012 First SIGNAL; Maemondo 2010 NEJ002; Mitsudomi 2010 WJTOG3405; Mok 2009 IPASS; Yang 2014) and one phase II study (Lou 2014) compared gefitinib with first‐line chemotherapy. IPASS compared gefitinib with carboplatin‐paclitaxel, but in Asian patients with adenocarcinoma who were light or never‐smokers (Mok 2009 IPASS). Maemondo 2010 NEJ002 randomised Asian chemotherapy‐naive patients with EGFR mutations to receive gefitinib or carboplatin‐paclitaxel. WJTOG3405 compared gefitinib with cisplatin plus docetaxel in Asian patients with EGFR mutations (Mitsudomi 2010 WJTOG3405). First‐SIGNAL compared first‐line gefitinib with gemcitabine plus cisplatin in Asian never‐smokers with lung adenocarcinoma (Han 2012 First SIGNAL). The phase III study by Yang 2014 compared first‐line pemetrexed and cisplatin followed by gefitinib maintenance therapy with gefitinib monotherapy alone in Asian non‐smoking patients. Patients were randomised at trial entry to either gefitinib or pemetrexed plus cisplatin chemotherapy. Patients in both arms then continued with maintenance gefitinib. Data were analysed in the intention‐to‐treat population and only data from the first phase of the study were included in this analysis. In the phase II study by Lou 2014, gefitinib was compared with carboplatin and paclitaxel in Asian patients who were either non‐smokers or light ex‐smokers.

We analysed a total of 2224 patients from the six studies in this group.

Second‐line studies

Three phase III studies (Lee 2010 ISTANA; Maruyama 2008 V‐15‐32; Sun 2012 KCSG‐LU08‐01) and three phase II studies (Dai 2013; Kim 2016; Li 2010) compared gefitinib with second‐line chemotherapy. ISTANA (Lee 2010 ISTANA), V‐15‐32 (Maruyama 2008 V‐15‐32) and the phase II study by Li 2010 included patients of Asian ethnicity but where mutation status was not always known, and compared gefitinib with docetaxel. KCSG‐LU08‐01 (Sun 2012 KCSG‐LU08‐01), Dai 2013 and Kim 2016 selected Asian patients with unknown EGFR status and compared gefitinib with second‐line pemetrexed. Secondary studies published by Sekine 2009 and Yamamoto 2010 conducted analyses on quality of life and disease control respectively in the V‐15‐32 trial.

We analysed a total of 1030 patients from the six studies in this group.

Maintenance studies

Two phase II studies compared the role of gefitinib as maintenance to chemotherapy. Ahn 2012 randomised Asian non‐smokers not progressing after first‐line pemetrexed‐cisplatin, to receive either gefitinib or pemetrexed ± cisplatin, in a two‐staged study design. Xu 2015 compared single‐agent pemetrexed with gefitinib in Asian patients not progressing after four to eight cycles of first‐line chemotherapy.

We analysed 258 patients in this group.

• EGFR mutation positive population (Comparison 6)

Nine studies were included in this group, six of which were first‐line studies (Crino 2008 INVITE; Han 2012 First SIGNAL; Maemondo 2010 NEJ002; Mitsudomi 2010 WJTOG3405; Mok 2009 IPASS; Yang 2014) and three of which were second‐line studies (Kim 2008 INTEREST; Maruyama 2008 V‐15‐32; Sun 2012 KCSG‐LU08‐01).

We included a total of 879 patients in this group.

Two phase III studies selected patients of Asian ethnicity who were also positive for EGFR mutations and compared gefitinib with first‐line carboplatin and paclitaxel or cisplatin and docetaxel respectively (Maemondo 2010 NEJ002; Mitsudomi 2010 WJTOG3405). In contrast, the IPASS (Mok 2009 IPASS) and First‐SIGNAL (Han 2012 First SIGNAL) studies selected Asian patients with adenocarcinomas, and conducted planned subgroup analyses on the EGFR mutation positive patients. IPASS compared first‐line gefitinib with carboplatin and paclitaxel and First‐SIGNAL compared gefitinib with gemcitabine and cisplatin. Yang 2014 conducted a post‐hoc analysis of EGFR mutation positive patients and compared first‐line pemetrexed and cisplatin followed by gefitinib maintenance with gefitinib alone. The INVITE phase II study in elderly patients that compared first‐line gefitinib with vinorelbine also conducted analysis of EGFR mutation positive patients but this study did not include any data that could be pooled (Crino 2008 INVITE).

We analysed a total of 802 patients in this group.

A further three phase III studies compared second‐line gefitinib with chemotherapy and conducted subgroup analyses in the EGFR mutation positive patients (Kim 2008 INTEREST; Maruyama 2008 V‐15‐32; Sun 2012 KCSG‐LU08‐01). INTEREST and V‐15‐32 compared gefitinib with docetaxel and KCSG‐LU08‐01 compared gefitinib with pemetrexed in this second‐line setting. The INTEREST study also analysed other biomarkers, such as EGFR gene copy number, EGFR protein expression and KRAS mutations, in addition to EGFR mutations. One study did not publish data that could be pooled (Maruyama 2008 V‐15‐32) and thus we included a total of 77 patients in this group.

3. Gefitinib at a specific dose versus a different dose (Comparison 7)

Three phase II studies compared the effect of two different doses of gefitinib, 250 mg and 500 mg in 527 patients (Fukuoka 2003 IDEAL I; Kris 2003 IDEAL II; Xue 2015). IDEAL I (Fukuoka 2003 IDEAL I) and IDEAL II (Kris 2003 IDEAL II) were multicentre, randomised, double‐blind, phase II studies that evaluated two doses of gefitinib (250 mg/day and 500 mg/day) as second‐ or third‐line therapy.

We analysed 431 patients in this group.

The third study randomised 96 patients who were stable after one month of gefitinib (250 mg/day) to either 250 mg/day or 500 mg/day as maintenance therapy (Xue 2015).

4. Gefitinib versus gefitinib combined with a chemotherapy regimen (Comparison 8)

Four studies compared gefitinib alone or in combination with chemotherapy. Two recently published studies examined the addition of chemotherapy to gefitinib versus gefitinib alone in the first‐line setting. A small study by An 2016 recruited 90 East Asian patients with an EGFR mutation and randomised them to receive gefitinib or gefitinib plus pemetrexed (500 mg/m²). In this study, pemetrexed or placebo was administered via intravenous infusion on day 1 of a 21‐day cycle. Gefitinib 250 mg was administered on days 2 to 16. A multicentre, phase II study by Cheng 2016 also compared gefitinib with and without pemetrexed as first‐line therapy. This study recruited 191 East Asian patients from China, Japan, Korea and Taiwan with advanced non‐squamous NSCLC with an activating EGFR mutation. Patients either received gefitinib 250 mg per day or gefitinib plus pemetrexed (500 mg/m²) infusion on day 1 of a 21‐day cycle.

We included a total of 281 patients in this group.

Chen 2007 compared 250 mg of daily oral gefitinib with gefitinib plus vinorelbine (15 mg/m²) every two weeks in 48 patients of Asian ethnicity with stage IV adenocarcinoma who had failed at least two lines of chemotherapy. Chen 2011 compared gefitinib alone with the combination of gefitinib plus tegafur (100 mg)/uracil (224 mg) in 115 Taiwanese patients with stage IIIB or IV adenocarcinoma who had failed first‐line chemotherapy.

We included a total of 163 patients in this group.

5. Gefitinib at any dose in combination with other chemotherapeutic agents versus the same chemotherapy agents alone (Comparison 9)

Five studies examined survival outcomes, objective response rates and toxicity (Giaccone 2004 INTACT I; Herbst 2004 INTACT II; Soria 2015 IMPRESS; Takeda 2010 WJTOG0203; Yu 2014). Overall, we included a total of 3110 patients.

INTACT I (Giaccone 2004 INTACT I) and INTACT II (Herbst 2004 INTACT II) were large, multicentre trials that examined the effect of the addition of two different doses of gefitinib to a chemotherapy regimen with the chemotherapy alone in chemotherapy‐naive patients. INTACT I compared the effect of the addition of gefitinib to a chemotherapy regimen that included gemcitabine and cisplatin and INTACT II a paclitaxel and carboplatin regime. WJTOG0203 compared the addition of 250 mg of gefitinib to platinum‐doublet chemotherapy in chemotherapy‐naive Japanese patients (Takeda 2010 WJTOG0203). In this study, patients were randomised to receive platinum doublet chemotherapy (Arm A) or platinum‐doublet chemotherapy for three cycles followed by gefitinib until disease progression (Arm B). The phase II study by Yu 2014 examined the addition of gefitinib to a first‐line pemetrexed and cisplatin chemotherapy schedule in Asian patients who were non‐smokers or light ex‐smokers.

In this group, we included 2845 patients.

The IMPRESS study was a phase III, multicentre study conducted across Europe and the Asia‐Pacific region (Soria 2015 IMPRESS). This study selected patients with EGFR mutation positive advanced NSCLC who had failed first‐line therapy with gefitinib. This study compared second‐line gefitinib plus chemotherapy (cisplatin and pemetrexed) with placebo plus the same chemotherapy regimen (cisplatin and pemetrexed). Two hundred and sixty‐five patients were included in this trial.

6. Gefitinib at any dose in combination with other chemotherapeutic agents versus a different combination of chemotherapeutic agent (Comparison 10)

No studies compared gefitinib in combination with a chemotherapeutic regime with a different regime of agents.

Data for all endpoints were not available in all published reports. A summary of efficacy and survival data is presented in Table 2.

Sequence generation

Seventeen of the 35 included studies reported adequate sequence generation (Ahn 2012; Cheng 2016; Crino 2008 INVITE; Cufer 2006 SIGN; Gaafar 2011 EORTC08021; Giaccone 2004 INTACT I; Goss 2009 INSTEP; Kim 2008 INTEREST; Maruyama 2008 V‐15‐32; Mitsudomi 2010 WJTOG3405; Mok 2009 IPASS; Morere 2010 IFCT‐0301; Soria 2015 IMPRESS; Sun 2012 KCSG‐LU08‐01; Thatcher 2005 ISEL; Yang 2014; Zhang 2012 INFORM). The remaining 18 studies were all described as randomised, but none provided any further information and so we classified them as having an uncertain risk of bias (An 2016; Chen 2007; Chen 2011; Dai 2013; Fukuoka 2003 IDEAL I; Han 2012 First SIGNAL; Herbst 2004 INTACT II; Kelly 2008 SWOG S0023; Kim 2016; Kris 2003 IDEAL II; Lee 2010 ISTANA; Li 2010; Lou 2014; Maemondo 2010 NEJ002; Takeda 2010 WJTOG0203; Xu 2015; Xue 2015; Yu 2014).

Allocation concealment

Allocation concealment was adequate in 11 of the included studies (Ahn 2012; Cufer 2006 SIGN; Gaafar 2011 EORTC08021; Goss 2009 INSTEP; Kim 2008 INTEREST; Mitsudomi 2010 WJTOG3405; Soria 2015 IMPRESS; Sun 2012 KCSG‐LU08‐01; Thatcher 2005 ISEL; Yang 2014; Zhang 2012 INFORM). Most of these studies used a minimisation method or centralised allocation procedure. The remaining studies did not report whether allocation was concealed and so are possibly at risk of bias.

Survival

See Analysis 1.1; Analysis 1.2; Analysis 1.3.

Four studies compared gefitinib with placebo in a general population (Gaafar 2011 EORTC08021; Goss 2009 INSTEP; Kelly 2008 SWOG S0023; Thatcher 2005 ISEL). The data presented examines the effect of gefitinib compared with placebo in the first‐line, second‐line and maintenance settings. Total pooling of data was not conducted for first‐ or second‐line therapy as only single studies were included. Pooling of data was only possible for maintenance treatment, as two studies were included (Gaafar 2011 EORTC08021; Kelly 2008 SWOG S0023). Gefitinib did not improve overall survival when compared with placebo, either when administered as first‐line (Goss 2009 INSTEP; hazard ratio (HR) 0.84, 95% confidence interval (CI) 0.62 to 1.14, P = 0.27), second‐line (Thatcher 2005 ISEL; HR 0.89, 95% CI 0.79 to 1.01, P = 0.06) or maintenance therapy (Gaafar 2011 EORTC08021; Kelly 2008 SWOG S0023; pooled HR 1.14, 95% CI 0.61 to 2.14, P = 0.69, I² = 85%, random‐effects model).

One‐year survival rates were improved by administration of gefitinib versus placebo as second‐line therapy (risk ratio (RR) 1.28, 95% CI 1.05 to 1.57, P = 0.02), but not as maintenance therapy (RR 0.90, 95% CI 0.78 to 1.04, P = 0.15). Progression‐free survival was not improved when gefitinib was compared with placebo as first‐line therapy and median progression‐free survival was reported as 1.4 months in both groups (HR 0.82, 95% CI 0.60 to 1.12, P = 0.21). Time to treatment failure was improved in favour of gefitinib as second‐line therapy, with a HR of 0.82 (95% CI 0.75 to 0.90, P < 0.0001): median progression‐free survival was 3 months with gefitinib, 2.6 months with placebo. Maintenance use of gefitinib after first‐line treatment improved progression‐free survival (HR 0.70, 95% CI 0.53 to 0.91, P = 0.007, I² = 32%).

Toxicity

See Analysis 1.4; Analysis 1.6.

We have pooled reported toxicity data from three studies in this comparison so as to examine the differences in toxicity between gefitinib and placebo or best supportive care (Gaafar 2011 EORTC08021; Goss 2009 INSTEP; Thatcher 2005 ISEL). Administration of gefitinib was significantly associated with Common Toxicity Criteria (CTC) grade 3 to 4 events such as skin rash (RR 7.92, 95% CI 1.46 to 43.03, P = 0.02, I² = 0%) and diarrhoea (RR 2.48, 95% CI 1.15 to 5.35, P = 0.02, I² = 0%). One study reported a statistically significant increase in alanine aminotransferase (ALT) with gefitinib (RR 9.11, 95% CI 1.18 to 70.32, P = 0.03). The risk of all other adverse events was either not estimable or not significantly different between the two groups.

Efficacy

See Analysis 1.22; Analysis 1.23.

Response was reported in only three of the four included studies (Gaafar 2011 EORTC08021; Goss 2009 INSTEP; Thatcher 2005 ISEL). We did not pool the data as the INSTEP study compared gefitinib with placebo as first‐line therapy, ISEL did so as second‐line therapy and the EORTC08021 trial as maintenance therapy. As first‐line therapy, gefitinib did not improve the overall response rate (RR 6.06, 95% CI 0.74 to 49.43, P = 0.09) or the disease control rate (RR 1.36, 95% CI 0.86 to 2.16, P = 0.19). This was reported as an overall response rate of 6% and 1% in the gefitinib and placebo groups, respectively, and the disease control rate was 31% and 23%, respectively. As second‐line therapy, the overall response rate was higher for gefitinib‐treated cases than for placebo (RR 6.42, 95% CI 2.82 to 14.64, P < 0.00001) and the disease control rate was also significantly higher for gefitinib (RR 1.24, 95% CI 1.06 to 1.44, P = 0.006). The overall response rate was 8% in the gefitinib group and 1% in the placebo group, and the disease control rate was 40% and 32%, respectively. Similarly, gefitinib improved the overall response rate and the disease control rate when used as maintenance therapy (RR 10.12, 95% CI 1.32 to 77.33, P = 0.03; RR 1.21, 95% CI 1.00 to 1.46, P = 0.05, respectively).

Quality of life and symptom improvement scores

Thatcher 2005 ISEL reported that the addition of gefitinib to "best supportive care" produced no significant changes in the quality of life subscale of the Functional Assessment of Cancer Therapy‐Lung (FACT‐L) questionnaire when compared with best supportive care alone. Gefitinib was associated with a statistically significant improvement in the symptom score (mean change from baseline ‐0.86 to ‐1.38; P = 0.019), but this did not meet predefined criteria. As described by Cella 2002, for changes in disease‐related symptoms to be classed as clinically relevant, the score must increase by two points. Goss 2009 INSTEP reported improvements in FACT‐L quality of life, FACT‐L Trial Outcome Index (TOI), lung cancer subscale (LCS) and Pulmonary Symptom Index (PSI) that were statistically non‐significant.

Survival

See Analysis 4.1; Analysis 4.2; Analysis 4.3.

Toxicity

See Analysis 4.4; Analysis 4.5; Analysis 4.6; Analysis 4.7; Analysis 4.8; Analysis 4.9; Analysis 4.10; Analysis 4.11.

We combined data to compare the toxicity profile of gefitinib with chemotherapy for first‐ and second‐line therapy to assess the overall effect in both groups. Data from Cufer 2006 SIGN, Crino 2008 INVITE, Kim 2008 INTEREST and Morere 2010 IFCT‐0301 were included. Gefitinib was generally better tolerated than chemotherapy. Gefitinib was associated with an increased risk of skin rash when compared with chemotherapy (RR 2.40, 95% CI 1.08 to 5.31, P = 0.03, I² = 4.7%, high quality of evidence). Gefitinib was associated with a decreased risk of constipation (RR 0.41, 95% CI 0.17 to 0.97, P = 0.04, I² = 0%, high quality of evidence), fatigue (RR 0.16, 95% CI 0.03 to 0.88, P = 0.04, I² = 8.2%, moderate quality of evidence), asthenia (RR 0.51, 95% CI 0.35 to 0.75, P = 0.0007, I² = 0%, high quality of evidence), neurotoxicity (RR 0.07, 95% CI 0.01 to 0.34, P = 0.001, I² = 0%, high quality of evidence), neutropenia (RR 0.04, 95% CI 0.02 to 0.06, P < 0.00001, I² = 43.1%, high quality of evidence), leukopenia (RR 0.03, 95% CI 0.00 to 0.22, P = 0.0005, I² = 0%, high quality of evidence) and febrile neutropenia (RR 0.12, 95% CI 0.06 to 0.23, P < 0.00001, I² = 0%, high quality of evidence). There were no differences between groups for any other measured adverse side effects including pruritus, diarrhoea, vomiting, anorexia, stomatitis, arthralgia, peripheral oedema, respiratory tract infection, dyspnoea, cough, anaemia, thrombocytopenia, hypokalaemia or pyrexia.

We assessed most of the toxicity outcomes as high‐quality evidence. We downgraded one outcome, fatigue, to a moderate quality of evidence as the study by Crino 2008 INVITE enrolled only 190 patients who were older than 70 years old, thus there was a risk of serious indirectness.

Efficacy

See Analysis 4.26; Analysis 4.27.

Only one first‐line study presented data on disease control rates and there was no reported improvement when administering gefitinib versus vinorelbine (RR 0.82, 95% CI 0.61 to 1.10, P = 0.19) (Crino 2008 INVITE). Disease control rates were 43.3% and 53.5% for gefitinib and chemotherapy, respectively. Two second‐line studies reported efficacy data (Cufer 2006 SIGN; Kim 2008 INTEREST). Pooled data showed that there was no improvement in overall response rate when comparing gefitinib and docetaxel as second‐line therapy (RR 1.16, 95% CI 0.85 to 1.59, P = 0.35, I² = 0%). Overall response rates were 9% to 13% for both the gefitinib and chemotherapy groups.

Quality of life and symptom improvement scores

See Analysis 4.28; Analysis 4.29; Analysis 4.30; Analysis 4.31.

We pooled data from the INVITE (Crino 2008 INVITE) and INTEREST (Kim 2008 INTEREST) studies. Patients who received gefitinib reported statistically significant improvements in quality of life as assessed by scores on the FACT‐L (standardised mean difference (SMD) 10.50, 95% CI 9.55 to 11.45, P < 0.00001, I² = 21%), LCS (SMD 3.63, 95% CI 3.08 to 4.19, P < 0.00001, I² = 0%) and TOI (SMD 9.87, 95% CI 1.26 to 18.48, P = 0.02, I² = 59%). One study also described an improvement in PSI scores (SMD 5.60, 95% CI 3.55 to 7.65, P < 0.00001) in patients who received gefitinib (Crino 2008 INVITE).

Survival

See Analysis 5.1; Analysis 5.2; Analysis 5.3; Analysis 5.4; Analysis 5.5; Analysis 5.6; Analysis 5.7.

Gefitinib versus first‐line chemotherapy

Five phase III studies compared gefitinib with first‐line platinum doublet chemotherapy (Han 2012 First SIGNAL; Maemondo 2010 NEJ002; Mitsudomi 2010 WJTOG3405; Mok 2009 IPASS; Yang 2014). The IPASS (Mok 2009 IPASS) and NEJ002 (Maemondo 2010 NEJ002) studies compared gefitinib with carboplatin‐paclitaxel. The WJTOG3405 study compared gefitinib with cisplatin‐docetaxel (Mitsudomi 2010 WJTOG3405). The First‐SIGNAL study compared gefitinib with gemcitabine‐cisplatin (Han 2012 First SIGNAL). The study by Yang 2014 compared gefitinib monotherapy with pemetrexed‐cisplatin followed by gefitinib maintenance.

Pooled analysis showed that gefitinib did not improve overall survival (HR 0.94, 95% CI 0.82 to 1.06, P = 0.31, I² = 0%) or the one‐year survival rate (RR 1.03, 95% C 0.97 to 1.09, P = 0.33, I² = 1%). One study reported median overall survival as 22 months in both groups. Progression‐free survival was higher in the gefitinib group than in the chemotherapy group (HR 0.65, 95% CI 0.43 to 0.98, P = 0.04, I² = 93%). Median progression‐free survival ranged from 5.5 to 6.4 months with chemotherapy to 5.7 to 10.4 months with gefitinib. Please refer to Figure 3 for the pooled progression‐free survival data from first‐line studies that included Asian patients.

Figure 3

---

---

Progression‐free survival: Gefitinib versus first‐line chemotherapy in an Asian population (Analysis 5.4).

Gefitinib versus second‐line chemotherapy

Two phase III studies compared gefitinib with second‐line docetaxel in patients of Asian ethnicity (Lee 2010 ISTANA; Maruyama 2008 V‐15‐32) and one phase III study compared gefitinib with pemetrexed (Sun 2012 KCSG‐LU08‐01). In pooled analysis of these three trials, there was no benefit on either overall survival or the one‐year survival rate for gefitinib over second‐line chemotherapy (HR 0.94, 95% CI 0.79 to 1.12, P = 0.50, I² = 0%; RR 0.94, 95% CI 0.81 to 1.11, P = 0.48, I² = 0%, respectively). Progression‐free survival was prolonged (HR 0.71, 95% CI 0.57 to 0.88, P = 0.002, I² = 40%; see Figure 4) in favour of gefitinib. Median progression‐free survival was 2 to 6.8 months with second‐line chemotherapy, and 2 to 10 months with gefitinib in the second‐line setting.

Figure 4

---

---

Progression‐free survival: Gefitinib versus second‐line chemotherapy in an Asian population (Analysis 5.5).

Toxicity

See Analysis 5.8; Analysis 5.9; Analysis 5.10; Analysis 5.11; Analysis 5.12; Analysis 5.13; Analysis 5.14; Analysis 5.15; Analysis 5.16; Analysis 5.17; Analysis 5.18; Analysis 5.19; Analysis 5.20; Analysis 5.21; Analysis 5.22; Analysis 5.23.

Gefitinib was generally well tolerated in this population. We pooled toxicity data from all studies. Compared to chemotherapy, the gefitinib group reported fewer adverse side effects such as nausea (RR 0.34, 95% CI 0.17 to 0.64, P = 0.001, I² = 0%), vomiting (RR 0.19, 95% CI 0.05 to 0.77, P = 0.02, I² = 56%, random‐effects model), anorexia (RR 0.36, 95% CI 0.27 to 0.49, P < 0.00001, I² = 18%), fatigue (RR 0.32, 95% CI 0.22 to 0.46, P < 0.00001, I² = 50%), arthralgia (RR 0.14, 95% CI 0.03 to 0.61, P = 0.009, I² = 0%), asthenia (RR 0.22, 95% CI 0.08 to 0.58, P = 0.002, I² = 13%), neurotoxicity (RR 0.07, 95% CI 0.02 to 0.24, P < 0.0001, I² = 0%), neutropenia (RR 0.11, 95% CI 0.05 to 0.27, P < 0.00001, I² = 82%, random‐effects model), anaemia (RR 0.18, 95% CI 0.12 to 0.29, P < 0.00001, I² = 4%), leukopenia (RR 0.07, 95% CI 0.02 to 0.23, P < 0.00001, I² = 77%, random‐effects model), thrombocytopaenia (RR 0.32, 95% CI 0.14 to 0.72, P = 0.006, I² = 22%) and febrile neutropenia (RR 0.09, 95% CI 0.03 to 0.28, P < 0.0001, I² = 0%). Other side effects were seen more frequently in the gefitinib group. Skin rash (RR 3.11, 95% CI 1.28 to 7.55, P = 0.01, I² = 60%, random‐effects model), diarrhoea (RR 2.79, 95% CI 1.57 to 4.94, P = 0.0005, I² = 0%), increased alanine aminotransferase (ALT) (RR 10.03, 95% CI 5.23 to 19.26, P < 0.00001, I² = 37%) and increased aspartate transaminase (AST) (RR 7.73, 95% CI 2.78 to 21.46, P < 0.0001, I² = 0%) were more frequent in gefitinib‐treated cases.

Efficacy

See Analysis 5.24; Analysis 5.25; Analysis 5.26.

Objective response rates were higher in the gefitinib group when compared with first‐line chemotherapy (RR 1.43, 95% CI 1.13 to 1.82, P = 0.003, I² = 76%, random‐effects model). The overall response rate ranged from 43% to 62.1% in the gefitinib group and 30.7% to 32.2% in the chemotherapy group. There was no effect on the disease control rate (RR 0.99, 95% CI 0.86 to 1.13, P = 0.86, I² = 80%, random‐effects model): 73% to 94% and 78% to 81%, respectively.

The overall response rate was not significantly improved in the gefitinib group compared with second‐line chemotherapy (RR 1.43, 95% CI 0.92 to 2.22, P = 0.11, I² = 46%). Two studies found that overall response rates were poor overall, but the gefitinib group performed better (23% to 28%) than the second‐line chemotherapy group (8% to 13%) (Lee 2010 ISTANA; Maruyama 2008 V‐15‐32). The disease control rate (RR 0.99, 95% CI 0.78 to 1.25, P = 0.92, I² = 46%) was statistically similar for both groups (34% and 33%, respectively).

Pooled data from two maintenance studies found that gefitinib improved the stable disease rate and the disease control rate (RR 0.64, 95% CI 0.44 to 0.93, P = 0.02; RR 0.65, 95% CI 0.49 to 0.85, P = 0.002, respectively). There was no improvement in the overall response rate with maintenance gefitinib (RR 0.88, 95% CI 0.41 to 1.87, P = 0.06, I² = 73%, random‐effects model).

Quality of life and symptom improvement scores

Three studies explored the impact of gefitinib versus chemotherapy on quality of life, but unfortunately the data could not be pooled (Lee 2010 ISTANA; Maruyama 2008 V‐15‐32; Mok 2009 IPASS). All three studies reported significantly improved quality of life in patients who received gefitinib as measured by the Trial Outcome Index (TOI). Mok 2009 IPASS and Maruyama 2008 V‐15‐32 stated that improvements as measured by FACT‐L were significant, but none recorded significant improvements on the lung cancer subscale (LCS).

Survival

See Analysis 6.1; Analysis 6.2; Analysis 6.3; Analysis 6.4.

Gefitinib versus first‐line chemotherapy

Five studies compared gefitinib with first‐line chemotherapy. Two of these studies selected patients with EGFR mutations (Maemondo 2010 NEJ002; Mitsudomi 2010 WJTOG3405), and the others selected patients based on clinical features and conducted subgroup analyses on patients positive for EGFR mutations (Han 2012 First SIGNAL; Mok 2009 IPASS; Yang 2014). We have separately analysed studies that selected EGFR mutants and those that selected patients based on clinical features then conducted subgroup analyses and progression‐free survival results are presented in Figure 5.

Figure 5

---

---

Progression‐free survival: Gefitinib versus first‐line chemotherapy in an EGFR mutation positive population (Analysis 6.3).

The two biomarker driven studies did not show any improvement in overall survival (HR 0.98, 95% CI 0.72 to 1.33, P = 0.90, I² = 54%). Progression‐free survival was significantly increased with gefitinib compared with first‐line chemotherapy (HR 0.39, 95% CI 0.26 to 0.59, P < 0.00001, I² = 66%, random‐effects model).

Three phase III studies conducted subgroup analyses in EGFR mutation positive patients. There was no improvement in overall survival (HR 0.95, 95% CI 0.68 to 1.33, P = 0.75, I² = 20%). However, there was a statistically significant improvement in progression‐free survival (HR 0.53, 95% CI 0.41 to 0.70, P < 0.00001, I² = 9%).

Pooled analysis of all first‐line studies that examined EGFR mutation positive patients showed that there was no difference in overall survival (HR 0.97, 95% CI 0.77 to 1.21, P = 0.76, I² = 15%). However, pooled data from these five studies showed that gefitinib was able to prolong progression‐free survival when compared with first‐line chemotherapy (HR 0.47, 95% CI 0.36 to 0.61, P < 0.00001, I² = 53%, random‐effects model), with median progression‐free survival improving from 5.5 to 6.3 months in the chemotherapy group to 9.2 to 10.4 months in the gefitinib group.

Gefitinib versus second‐line chemotherapy

When comparing gefitinib with second‐line chemotherapy, data were available from two studies (Kim 2008 INTEREST; Sun 2012 KCSG‐LU08‐01). This showed that gefitinib did not improve overall survival (HR 0.83, 95% CI 0.41 to 1.66, P = 0.60). There was a statistically significant improvement in progression‐free survival (HR 0.24, 95% CI 0.12 to 0.47, P < 0.0001, I² = 0%) in EGFR mutation positive patients. Progression‐free survival for this analysis is presented in Figure 6.

Figure 6

---

---

Progression‐free survival: Gefitinib versus second‐line chemotherapy in an EGFR mutation positive population (Analysis 6.4).

Efficacy

See Analysis 6.5; Analysis 6.6; Analysis 6.7.

Survival

See Analysis 7.1.

Two multicentre, randomised, double‐blind, phase II studies evaluated differing doses of gefitinib (250 mg and 500 mg) in the second‐line setting (Fukuoka 2003 IDEAL I; Kris 2003 IDEAL II). There was no significant effect on one‐year survival (RR 0.83, 95% CI 0.61 to 1.11, P = 0.21, I² = 0%). HRs were not available for meta‐analysis. Median overall survival ranged from 7 to 7.6 months in patients given 250 mg, and 6 to 8 months in those given 500 mg of gefitinib. Median progression‐free survival ranged from 2.7 to 7 months and from 2.8 to 6 months in patients given 250 mg and 500 mg, respectively.

One study examined the effect of a higher dose of gefitinib in patients that had been stable after one month of 250 mg/day dosing of gefitinib (Xue 2015). In this study, there was no difference in progression‐free or overall survival with a higher dose of gefitinib (500 mg/day versus 250 mg/day: median progression‐free survival 5.30 months versus 6.23 months, P = 0.167; median overall survival 13.70 months versus 18.87 months, P = 0.156).

Toxicity

See Analysis 7.2; Analysis 7.3; Analysis 7.4; Analysis 7.5; Analysis 7.6; Analysis 7.7; Analysis 7.8; Analysis 7.9.

Data from all three studies were available for pooling (Fukuoka 2003 IDEAL I; Kris 2003 IDEAL II; Xue 2015). A gefitinib dose of 500 mg had a marginally worse toxicity profile when compared with the lower dose of 250 mg. This higher dose was associated with an increased rate of diarrhoea (RR 8.36, 95% CI 1.58 to 44.34, P = 0.01, I² = 0%) and skin rash (RR 8.13, 95% CI 1.51 to 43.72, P = 0.01, I² = 0%). Other reported side effects such as pruritus, acne, vomiting, anorexia, asthenia, neutropenia, leukopenia and dyspnoea were not significantly different between doses.

Efficacy

See Analysis 7.10; Analysis 7.11.

Pooled analysis of two studies found no significant difference in overall response rate (RR 0.92, 95% CI 0.58 to 1.46, P = 0.72, I² = 0%) between doses (Fukuoka 2003 IDEAL I; Kris 2003 IDEAL II). Overall response rates in the 250mg arm were reported as 18% and 12% in the IDEAL I and IDEAL II trials respectively, compared with ORR rates of 19% and 9% respectively, in patients receiving 500mg of gefitinib. Complete and partial response rates were only reported individually in the IDEAL I paper, and were 10% and 18.1%, respectively.

A higher dose of gefitinib as maintenance treatment did not improve the overall response rate (12.5% versus 12.5%, P = 1) (Xue 2015).

Quality of life and symptom improvement scores

See Analysis 7.12; Analysis 7.13.

Two studies reported changes in quality of life and symptom improvement scores (Fukuoka 2003 IDEAL I; Kris 2003 IDEAL II).

Quality of life improvements were also measured using the Trial Outcome Index (TOI), a summary score of the physical and functional domains of FACT‐L and the lung cancer subscale (a validated subscale of the FACT‐L questionnaire). No statistically significant difference was found between 250 mg and 500 mg of gefitinib in the rate of change of the FACT‐L and TOI scales (SMD 3.70, 95% CI ‐7.28 to 14.69; P = 0.51, I² = 0% and SMD 7.38, 95% CI ‐2.30 to 17.05; P = 0.14, I² = 0%, respectively). Unfortunately, extractable data from the published papers were inconsistently reported and thus not all data were pooled for analysis.

Data from the IDEAL II study further correlated symptom improvement with objective response and survival. When given a dose of 250 mg of gefitinib, all patients who experienced a partial response also experienced symptom improvement. Patients with stable or progressive disease who experienced symptom improvement also had a longer median survival time compared to those in the same tumour progression category without symptom improvement.

Survival 

See Analysis 8.1; Analysis 8.2; Analysis 8.3.

In the first‐line setting, two studies compared gefitinib alone with gefitinib plus pemetrexed (An 2016; Cheng 2016). One study reported no difference in median survival between the gefitinib and gefitinib plus chemotherapy group (32 months versus 34 months respectively) (An 2016). The other study did not present survival data (Cheng 2016). There was, however, a statistically significant improvement in progression‐free survival in favour of gefitinib plus chemotherapy over gefitinib alone (HR 0.69, 95% CI 0.49 to 0.96; P = 0.03; median progression‐free survival 12.6 months versus 18.3 months) (Cheng 2016).

In the second‐line or greater setting, median overall survival improved from 13.3 months (Chen 2007) and 18.3 months (Chen 2011) to 23.4 months and 23.6 months, respectively. This improvement was not statistically significant. Combining gefitinib with either vinorelbine or tegafur/uracil did not improve the one‐year survival rate (RR 1.15, 95% CI 0.92 to 1.43; P = 0.22; I² = 43%). Gefitinib plus chemotherapy improved one‐year progression‐free survival (RR 2.29, 95% CI 1.38 to 3.80; P = 0.001). However, the HR for progression‐free survival was only presented in Chen 2011 (HR 0.65, 95% CI 0.43 to 0.97; P = 0.04: median progression‐free survival improved from 7.1 months (Chen 2007) and 5.3 months (Chen 2011) to 12.8 months and 8.3 months, respectively).

Toxicity

See Analysis 8.4; Analysis 8.5; Analysis 8.6; Analysis 8.7; Analysis 8.8; Analysis 8.9; Analysis 8.10; Analysis 8.11; Analysis 8.12; Analysis 8.13; Analysis 8.14; Analysis 8.15.

We pooled toxicity data from three studies (An 2016; Chen 2007; Cheng 2016). Both regimens were well tolerated with no significant difference in rates of skin rash, diarrhoea, constipation, fatigue, blood counts, nausea or vomiting.

Pooled data from both first‐line studies did show that the addition of pemetrexed chemotherapy to gefitinib resulted in higher rates of raised ALT (RR 2.57, 95% CI 1.09 to 6.04; P = 0.03; I² = 63%) but not AST (RR 1.47, 95% CI 0.56 to 3.88; P = 0.44; I² = 0%).

Efficacy

See Analysis 8.16; Analysis 8.17; Analysis 8.18.

When comparing gefitinib alone to gefitinib plus chemotherapy as first‐line therapy, there was no improvement in overall response rate (RR 1.02, 95% CI 0.89 to 1.17; P = 0.73; I² = 26%) or rate of stable disease (RR 0.67, 95% CI 0.39 to 1.16; P = 0.16; I² = 0%).

In the second‐line setting, the addition of chemotherapy to gefitinib did not result in an improvement in either partial radiological response (RR 1.02, 95% CI 0.71 to 1.47; P = 0.92; I² = 0%) or stable disease (RR 1.30, 95% CI 0.84 to 2.03; P = 0.24; I² = 16%).

Survival

See Analysis 9.1; Analysis 9.2; Analysis 9.3.

Meta‐analysis of two phase II, first‐line trials examining 1411 patients showed that the addition of gefitinib (250 mg/day) to a chemotherapy regimen in chemotherapy‐naive patients did not change the one‐year survival rate (RR 0.95, 95% CI 0.84 to 1.08, P = 0.44, I² = 0%) (Giaccone 2004 INTACT I; Herbst 2004 INTACT II).

Two trials compared the addition of first‐line gefitinib to chemotherapy with chemotherapy alone in Asian patients only (Takeda 2010 WJTOG0203; Yu 2014). There was no improvement in overall survival (HR 0.86, 95% CI 0.72 to 1.02, P = 0.08, I² = 0%), however there was a statistically significant improvement in progression‐free survival (HR 0.69, 95% CI 0.62 to 0.77, P < 0.00001, I² = 18%).

A single phase III trial recruited only EGFR mutation positive patients who had failed prior first‐line gefitinib, and the addition of gefitinib to chemotherapy did not improve progression‐free survival (HR 0.86, 95% CI 0.65 to 1.13, P = 0.28) (Soria 2015 IMPRESS). Overall survival appeared to be better in the chemotherapy alone group (HR 1.62, 95% CI 1.05 to 2.50, P = 0.03).

Toxicity

See Analysis 9.4; Analysis 9.5; Analysis 9.6.

Pooled data from all five trials showed that the addition of gefitinib to a chemotherapeutic regimen resulted in increased rates of skin rash (RR 2.98, 95% CI 1.54 to 5.77, P = 0.001, I² = 28%), acne (RR 4.95, 95% CI 1.09 to 22.51, P = 0.04, I² = 0%) and diarrhoea (RR 2.04, 95% CI 1.17 to 3.58, P = 0.01, I² = 17%). Other measured side effects such as pruritus, vomiting, nausea, anorexia, asthenia, dyspnoea, anaemia, neutropenia and leukopenia were not significantly increased.

Efficacy

See Analysis 9.16.

In the first‐line setting, the addition of gefitinib to chemotherapy did not effect the overall response rate in either the unselected population (RR 1.07, 95% CI 0.94 to 1.22, P = 0.28, I² = 0%) or the Asian population (RR 1.14, 95% CI 0.93 to 1.40, P = 0.20, I² = 0%). The overall response rate ranged from 30% to 51% in the gefitinib plus chemotherapy group and 29% to 50% in the chemotherapy group.

There was also no improvement in the overall response rate in the second‐line setting (RR 0.93, 95% CI 0.66 to 1.31, P = 0.66, I² = 0%), and the overall response rate was 32% in the gefitinib plus chemotherapy group and 34% in the chemotherapy alone group.

Quality of life and symptom improvement scores

In the first‐line setting, the WJTOG0203 study reported a disease‐related symptoms assessment (Takeda 2010 WJTOG0203). Sequential administration of gefitinib was reported to provide better symptom control, however these differences were not statistically significant. The adjusted mean of initial summed scores of the lung cancer subscale were 20.3 for Arm A and 20.6 for Arm B. The adjusted lung cancer subscale scores at 12 and 18 weeks were 21 and 20.9 for Arm A and 21.8 and 21.2 for Arm B, respectively.

In the second‐line setting, the IMPRESS study reported that the improvements in quality of life were no different when gefitinib plus chemotherapy was compared to placebo plus chemotherapy as measured by the Trial Outcome Index (TOI) (29% versus 30.2%, respectively), FACT‐L (35.5% versus 38%, respectively) or lung cancer subscale (43.5% versus 42.6%, respectively) (Soria 2015 IMPRESS). There was also no difference in the time to worsening of health‐related quality of life as measured by the TOI, FACT‐L and lung cancer subscale.

These data could not be pooled for meta‐analysis.