The role of the epidermal growth factor receptor tyrosine kinase inhibitors as therapy for advanced, metastatic, and recurrent non-small-cell lung cancer: a Canadian national consensus statement

Purpose To provide consensus recommendations on the use of epidermal growth factor receptor tyrosine kinase inhibitors (egfr-tkis) in patients with advanced or meta-static non-small-cell lung cancer (nsclc). Methods Using a systematic literature search, phase ii trials, randomized phase iii trials, and meta-analyses were identified for inclusion. Results A total of forty-six trials were included. Clear evidence is available that egfr-tkis should not be administered concurrently with platinum-based chemotherapy as first-line therapy in advanced or metastatic nsclc. Evidence is currently insufficient to recommend single-agent egfr-tkis as first-line therapy either in unselected populations or in populations selected on the basis of molecular or clinical characteristics. Following failure of platinum-based chemotherapy, the evidence suggests that second-line egfr-tkis or second-line chemotherapy result in similar survival. Quality of life and symptom improvement for patients treated with an egfr-tki appear better than they do for patients treated with second-line docetaxel. Sequence of therapy may not appear to be important, but if survival is the outcome of interest, the goal should be to optimize the number of patients receiving three lines of therapy. Based on available data, molecular markers and clinical characteristics do not appear to be predictive of a differential survival benefit from an egfr-tki and therefore those factors should not be used to select patients for egfr-tki therapy. Conclusions The egfr-tkis represent an additional therapy in the treatment of advanced or metastatic nsclc. The results of ongoing clinical trials may define the optimal role for these agents and the effectiveness of combinations of these agents with other targeted agents.


INTRODUCTION
Lung cancer represents a major health burden in Canada. Approximately 23,300 new lung cancer cases and 19,900 deaths from lung cancer occurred in 2007, most of which were non-small-cell lung cancer (NSCLC) 1 . Most of these patients either present with or develop metastatic disease at some point during their illness; potentially, they are candidates for systemic therapy approaches such as chemotherapy.
Until the late 1990s, therapeutic nihilism about the benefit of systemic chemotherapy in the treatment of advanced and metastatic NSCLC was widespread. Publication of the Non-small Cell Lung Cancer Collaborative Group meta-analysis in 1995 established the association of first-line platinum-based chemotherapy with a modest improvement in survival for patients with metastatic disease 2 . The introduction of newer drugs such as vinorelbine, gemcitabine, paclitaxel, and docetaxel have resulted in further small improvements, although most patients still experience disease progression within a short time, with a median time to progression (TTP) of approximately 4 months [3][4][5] .
At the time of progression following platinumbased chemotherapy, many patients maintain a good 28 28 28 28 28 performance status (PS) and may be candidates for further systemic therapy. Recent trials have established that second-line chemotherapy with docetaxel [6][7][8][9] improves survival and quality of life (QOL) as compared with best supportive care (BSC) and that survival of patients treated with docetaxel or pemetrexed is similar 10 . Guidelines for the management of NSCLC, including those from Cancer Care Ontario's Program in Evidence-Based Care (CCO-PEBC) 11 now recommend either of those agents as second-line chemotherapy options 11,12 . Despite these advancements in the treatment of NSCLC, there is still a strong need for additional and better treatment options. Recently, a greater understanding of the molecular abnormalities associated with NSCLC has led to evaluation of new therapeutic targets for NSCLC. The epidermal growth factor receptor (EGFR) is one target commonly overexpressed in  . Early-phase clinical trials showed that EGFR tyrosine kinase inhibitors (TKIs) such as erlotinib and gefitinib had antitumour activity, and this finding prompted their further evaluation in advanced NSCLC 16 . These agents have been evaluated extensively in phase II and III trials over the last few years, confirming the promising activity seen in phase I trials, and the TKIs have been incorporated into treatment algorithms for patients after progression on standard chemotherapy options 11 .
Because of a favourable toxicity profile of the TKIs, many clinicians felt that it might be appropriate to expand their role in the treatment of advanced and metastatic NSCLC. A need therefore exists to clarify the role of EGFR-TKIs in the treatment of NSCLC. The present paper represents a consensus view of a representative sample of Canadian lung cancer medical oncologists on the role of EGFR-TKIs in the treatment of NSCLC based on a systematic review of currently available evidence.

MATERIALS AND METHODS
Medical oncologists specializing in thoracic oncology from five provinces across Canada were invited to participate in a consensus meeting. Six oncologists attended the consensus meeting, and three additional oncologists, plus one pathologist, provided input into the consensus process. Three key questions were identified to be addressed by the group: • What is the role of EGFR-TKIs as first-line therapy of advanced or metastatic NSCLC as a single agent or in combination with chemotherapy? • What is the role of EGFR-TKIs following progression after platinum-based chemotherapy (singleagent EGFR-TKI vs. BSC, EGFR-TKI vs. chemotherapy, and EGFR-TKI in combination with another agent)? • Do any patient subpopulations, or clinical and molecular characteristics, predict for additional benefit from EGFR-TKI therapy?

Literature Search
A search of the MEDLINE database for 2000-2007 was conducted using the terms "non-small-cell lung cancer," "epidermal growth factor receptor tyrosine kinase inhibitor," "erlotinib," and "gefitinib." The search excluded articles prior to 2000, because the EGFR-TKIs are new agents and their initial phase I trials were known to be conducted during the selected time period. Conference proceedings of the American Society of Clinical Oncology 2000-2007 and the International Association for the Study of Lung Cancer 2007 World Conference on Lung Cancer were also searched. Finally, the list of included articles was reviewed by the consensus panel for omissions.

Study Selection Criteria
Articles published as full reports or as abstracts and conference presentations were included if they focused on • EGFR-TKI alone or in combination with chemotherapy in the first-line setting, • EGFR-TKI as second-or third-line therapy following progression of platinum-based chemotherapy, or • clinical and molecular characteristics that may predict additional benefit from EGFR-TKI therapy.
The literature search results were reviewed by two authors (PE, FK), and articles that met the foregoing criteria were selected for retrieval. The outcomes of interest were overall survival (OS), time to disease progression, tumour response rate, molecular and clinical predictors of benefit from EGFR-TKI therapy, and QOL or symptom improvement. Single-arm phase II trials were included only if no data from randomized trials were available. Forty-three individual trials (eight phase III, eleven randomized phase II, and twenty-four single-agent phase II trials) met the eligibility criteria for the present consensus statement. Only studies published in English were considered.

External Review
Final consensus statement draft recommendations were distributed electronically to reviewers. The review panel consisted of practitioners who had attended the consensus meeting and others who were not in attendance. The comments resulting from this review were incorporated into the final document.

First-Line Treatment
What is the role of EGFR-TKIs as first-line therapy of advanced or metastatic NSCLC as a single agent or in combination with chemotherapy? EGFR-TKI therapy in the treatment of advanced or metastatic NSCLC. These recommendations apply both to unselected populations and to patients selected on the basis of activating mutations of the EGFR gene or of clinical characteristics predictive of higher response to therapy.
There is evidence of tumour response to singleagent EGFR-TKI as first-line therapy for advanced NSCLC. Response rates to EGFR-TKI therapy appear to be higher in patients selected on the basis of activating mutations of the EGFR gene.
Randomized trials are needed to evaluate the effect of first-line EGFR-TKI on survival.

What Is the Role of Single-Agent EGFR-TKIs in Patients with Adenocarcinoma with Bronchioloalveolar Features?
Key Evidence: The literature search identified a consensus document on systemic therapy of bronchioloalveolar carcinoma (BAC) 41 . It states that there is no evidence to confirm or refute the assertion that the sensitivity of BAC to chemotherapy is any different from that of other histologic subtypes of NSCLC.

Consensus Recommendation:
There is no evidence to suggest that BAC should be treated differently from other types of NSCLC. The evidence is currently insufficient to recommend EGFR-TKIs as first-line therapy for the treatment of BAC.

What Is the Role of First-Line EGFR-TKIs in Combination with Platinum-based Chemotherapy in Patients with NSCLC?
Key Evidence: Four large randomized trials evaluated EGFR-TKIs in combination with platinum-based chemotherapy in patients with good PS with stage III/IV NSCLC (n = 4348, Table III  CURRENT ONCOLOGY-VOLUME 16,NUMBER 1 ied between the trials; however, all four trials failed to demonstrate any improvement in response rate with the addition of an EGFR-TKI to platinum-based chemotherapy [46][47][48][49] . Time to worsening of symptoms did not differ significantly between the groups 46,47,49 . No differences were observed in time to disease progression or in median and 1-year survival between patients randomized to chemotherapy alone and those randomized to chemotherapy plus an EGFR-TKI 46-49 (see Table III).

Consensus Recommendation:
Clear evidence from four randomized trials shows that concurrent administration of an EGFR-TKI with first-line platinum-based chemotherapy does not prolong survival in unselected patients with NSCLC.

What Is the Role of Single-Agent EGFR-TKIs Compared with Chemotherapy in Chemonaïve Patients with NSCLC?
Key Evidence: Two randomized trials compared firstline therapy with an EGFR-TKI with chemotherapy in chemonaïve patients with stage III/IV NSCLC and PS 0-2 (n = 299, Table IV) 50,52 . Lilenbaum randomized patients with poor PS (score of 2) to treatment with either carboplatin and paclitaxel (carboplatin AUC 6 and paclitaxel 200 mg/m 2 for 4 cycles) or erlotinib 150 mg daily 52 ; Crinò randomized elderly patients (more than 70 years of age) to vinorelbine 30 mg/m 2 IV on days 1 and 8 of a 21-day cycle or gefitinib 250 mg daily 50 .
Lilenbaum observed a higher response rate among patients treated with chemotherapy than with erlotinib [overall response (OR): 12% vs. 2%; OR + stable disease (SD): 53% vs. 39%]. Additionally, patients randomized to carboplatin-paclitaxel had a longer time to progression (3.5 months vs. 1.9 months) and a greater survival (9.1 months vs. 6.6 months), although these differences were not statistically significant 52 FACT-L) and in LCS. Gefitinib appeared to be better tolerated than vinorelbine 50 .
A third trial evaluated various doses and schedules of erlotinib with carboplatin and paclitaxel 51 . No significant differences were observed among the three treatment groups (Table IV).

Consensus Recommendation:
The evidence is currently insufficient to recommend the use of an EGFR-TKI over chemotherapy in the first-line therapy of patients with NSCLC. Available evidence raises the possibility that survival of patients with poor PS treated with firstline EGFR-TKI may be less than that of patients treated with platinum-based chemotherapy.

Second-Line and Subsequent Treatment for Relapsed or Recurrent Disease
What is the role of EGFR-TKIs following progression after platinum-based chemotherapy (single-agent EGFR-TKI vs. BSC, EGFR-TKI vs. chemotherapy, and EGFR-TKI in combination with another agent)?

What Is the Role of EGFR-TKIs as Second-or Third-Line Therapy Following Progression of Platinum-based Chemotherapy?
Key Evidence: Two guidelines developed by CCO-PEBC, addressing the role of an EGFR-TKI as subsequent therapy for NSCLC, were identified 11,53 . Both documents recommend the use of erlotinib as second-or third-line therapy for NSCLC in patients who are not candidates for further chemotherapy. Four randomized phase II and III trials in PS 0-2 patients with stage III/IV NSCLC who were not considered candidates for further chemotherapy examined EGFR-TKIs as subsequent therapy following progression of platinum-based chemotherapy (n = 2849, Table  V). Two large phase III studies evaluated erlotinib 150 mg (BR.21) or gefitinib 250 mg [ISEL (Iressa Survival Evaluation in Lung Cancer)] daily compared with placebo 56,57 , and two randomized phase II studies [IDEAL 1 and 2 (Iressa Dose Evaluation in Advanced Lung Cancer 1 and 2)] compared two doses of gefitinib (250 or 500 mg daily) 54,55 . In the IDEAL 1 and 2 trials, no differences were observed in any outcomes examined between gefitinib 250 mg and 500 mg daily.
Results of the BR. 21 and ISEL trials demonstrated that erlotinib (2.2 months vs. 1.8 months) and gefitinib (3.0 months vs. 2.6 months) significantly prolong time to disease progression 56,57 . Statistically significant improvements were also seen in OS with erlotinib as compared with placebo (6.7 months vs. 4.7 months, p < 0.001) 56 , and a trend toward improved survival was observed with gefitinib (5.6 months vs. 5.1 months, p = 0.087) 57 .
In the BR.21 trial, patients receiving erlotinib experienced significantly longer time to deterioration in several lung cancer-related symptoms (cough, pain, dyspnea) and in overall physical function 58 . In the ISEL trial, a greater proportion of patients randomized to gefitinib experienced improvement in disease-related symptoms (27% vs. 22%). Similarly, patients randomized to gefitinib experienced a significantly greater improvement in LCS scores (-1.38 vs. -0.86, p = 0.019) 57 .

Consensus Recommendation:
In patients with advanced or metastatic NSCLC who are not candidates for further chemotherapy, the use of an EGFR-TKI (as 32 32 32 32 32    65,66 . However no differences were observed in response rate between gefitinib and docetaxel in the other two trials 59,62 . No significant differences were observed in TTP or OS in patients treated with gefitinib or docetaxel. In the trial by Niho et al., the proportion of patients randomized to docetaxel who received third-line EGFR-TKI therapy was greater than the proportion of patients randomized to gefitinib who received third-line chemotherapy. That trial did not meet its primary outcome of non-inferiority of gefitinib (upper limit of 95% CI ≤ 1.25) as compared with docetaxel (HR: 1.12; 95% CI: 0.89 to 1.40) 65,66 . However, the larger INTEREST trial (Iressa non-small-cell lung cancer trial evaluating response and survival against Taxotere) demonstrates that gefitinib was non-inferior to docetaxel (HR: 1.02; 95% CI: 0.905 to 1.15), in which the definition of non-inferiority accepted a CI going up to 1.154 62 . The proportion of patients receiving effective third-line therapy was similar between the two treatment arms in that trial.
Another four randomized phase II studies evaluated gefitinib 250 mg or erlotinib 150 mg daily with other agents (oral vandetanib 300 mg daily 60 ; bortezomib 1.6 mg/m 2 IV on days 1 and 8 of a 21-day cycle 64 ; vinorelbine 15 mg/m 2 IV on day 1, and gefitinib 250 mg daily on days 2-14 every 2 weeks 61 ; bevacizumab 15 mg IV on day 1 every 3weeks; docetaxel 75 mg/m 2 on day 1 of a 3-week cycle; pemetrexed 500 mg/m 2 on day 1 of a 3-week cycle) 63 either as single agents or in combination (n = 386, Table VI). No firm conclusions can be drawn from any of these trials, although compared with erlotinib alone, the combination of erlotinib plus bevacizumab demonstrated improvement in response rate (17.9% vs. 12.2%), TTP (4.4 months vs. 3.0 months), and OS (13.7 months vs. 8.6 months) 63 . A phase III trial of that combination is ongoing. Fully powered phase III trials are ongoing to compare gefitinib with vandetanib and to assess whether bevacizumab adds to the efficacy of single-agent erlotinib.

Consensus Recommendation:
The evidence suggests that second-line EGFR-TKI or second-line chemotherapy results in similar survival. Sequence does not appear to be important, but if survival is the outcome of interest, the goal should be to optimize the number of patients receiving three lines of effective therapy. The evidence is currently insufficient to recommend second-line therapy with a combination of an EGFR-TKI and another targeted agent. Ongoing randomized phase III trials are currently addressing these questions.

How Do QOL and Symptom Control Compare in Patients Treated with Chemotherapy as Compared with EGFR-TKIs?
Key Evidence: Two of the three trials that compared gefitinib and docetaxel also examined QOL and symptom improvement 59,62 .
In the SIGN trial (Second-Line Indication of Gefitinib in NSCLC), a greater proportion of patients randomized to gefitinib experienced symptom improvement as assessed by LCS (36.8% vs. 26%) and QOL improvement as assessed by the FACT-L (33.8% vs. 26%) 59 . In addition, in the INTEREST trial, significantly more patients randomized to the gefitinib arm showed improvements in FACT-L score (25.1% vs. 14.7%, p < 0.0001) and trial outcome index (17.3% vs. 10.3%, p = 0.0026). Symptom improvement rates were also better with gefitinib than with docetaxel, but this difference was not statistically significant 62 .
Key Recommendation: Symptom control and QOL appear to be better in patients treated with an EGFR-TKI than in those treated with either BSC or secondline chemotherapy with docetaxel. In decisions about treatment following failure of platinum-based chemotherapy, QOL and patient choice are important.

What Is the Role of Single-Agent EGFR-TKI Therapy in Previously Treated Patients with EGFR Gene Mutations or High Gene Copy, or EGFR Protein Expression?
Key Evidence: Four single-arm phase II trials evaluated gefitinib 250 mg daily in patient populations (n = 117) selected for the presence of activating mutations of the EGFR gene assessed by polymerase chain reaction (PCR) analysis or for high EGFR gene copy assessed using fluorescence in situ hybridization (FISH). Patients had stage III/IV disease and PS 0-2, and most had received prior chemotherapy. High response rates were observed (48%-90%) [67][68][69][70] . Time to disease progression ranged from 6.4 months to 12.9 months, with a median survival of 15.5 months reported in one study 69 . Given that EGFR mutations are thought to represent a favourable prognostic factor, the significance of these data are unclear, and randomized trials   CURRENT ONCOLOGY-VOLUME 16,NUMBER 1 are needed to determine if the survival of patients with EGFR mutations or high EGFR gene copy treated with an EGFR-TKI is superior to that of similar patients treated with second-line chemotherapy.

Consensus Recommendations:
There is evidence that patients with previously treated NSCLC and EGFR mutations or increased EGFR gene copy respond to an EGFR-TKI. However, the evidence is insufficient at this time to select patients for EGFR-TKI therapy rather than for second-line chemotherapy based on any EGFR marker.

Clinical and Molecular Predictors of Benefit
Do any patient subpopulations, or clinical and molecular characteristics, predict for additional benefit from EGFR-TKI therapy?  57 .  76 .

Clinical Predictors of Survival with an EGFR-TKI:
No data were available concerning clinical predictors of survival from the INTACT (Iressa NSCLC Trial Assessing Combination Treatment) 1 and 2 trials 80 . In a subset analysis of never smokers (n = 113) from the TRIBUTE (Tarceva Responses in Conjunction with Paclitaxel and Carboplatin) trial, a significant improvement in survival was observed from the addition of erlotinib (HR: 0.49; 95% CI: 0.28 to 0.85) 81 . Similar findings were observed in TALENT (Tarceva Lung Cancer Investigation Trial). Improved OS and PFS were observed for patients receiving erlotinib who had never smoked (HR: 0.39; p = 0.25), although this interaction did not achieve statistical significance 74,75 .
In contrast, subgroup analyses from the INTEREST trial comparing gefitinib with docetaxel suggest that these clinical variables do not predict a differential benefit for an EGFR-TKI over chemotherapy. There was no difference in the survival of patients with adenocarcinoma histology, never smokers, Asian ethnicity, and female sex when treated with either gefitinib or docetaxel 79 .

Molecular Predictors of Response to an EGFR-TKI:
The predictive value of various molecular abnormalities have been examined in the randomized trials included in the present consensus document. These include mutations of the EGFR gene, increased EGFR gene copy assessed by FISH or EGFR amplification assessed by quantitative PCR, EGFR expression [by immunohistochemistry (IHC)], and mutations of the KRAS gene. Table IX summarizes predictors of response.
The presence of an activating mutation of the EGFR gene is associated with an increased likelihood of response to single-agent EGFR-TKI. Analyses of tumour samples from the IDEAL 1 and 2 trials (n = 425) evaluating gefitinib monotherapy demonstrated that patients whose tumour had an EGFR mutation had a better OR with gefitinib than did patients lacking the mutation (n = 79: 46% vs.10%, p = 0.005) 80 . In the BR.21 (n = 177: 15.8% vs. 7.4%, p = 0.35) and ISEL trials (n = 215: 37.5% vs. 2.6%), the presence of an EGFR mutation was associated with a nonsignificant increase in response rate. In BR.21, when only exon 19 deletion and L858R mutations were considered, the difference in response rate as compared with wild-type EGFR or other mutations was significant (27% vs. 7%, p = 0.035) 85 . The subset analysis of tumour samples from the IN-TACT 1 and 2 trials evaluating the addition of gefitinib to standard first-line chemotherapies demonstrated that patients whose tumours had an EGFR mutation had a higher response to chemotherapy plus gefitinib than did those without a mutation (n = 170: 72% vs. 55%, p = 0.2) 80 . Similar findings were observed in the TRIBUTE trial for patients with EGFR mutations (n = 228: 53% vs. 21%, p < 0.01) 82,84 , but no statistically significant correlation was observed between response rates and mutation status in the TALENT trial 74,75 . Increased EGFR gene copy or EGFR amplification also appears to be associated with an increased response rate to single-agent EGFR-TKI. The IDEAL 1 and 2 trials demonstrated that EGFR amplification was associated with a higher response to gefitinib than was seen with tumours without EGFR amplification; however, this difference was not statistically significant (n = 90: risk ratio: 29% vs.15%; p = 0.319). Patients with an EGFR mutation or gene amplification had a significantly improved response rate as compared with patients with neither EGFR amplification nor mutation (60% vs. 10%, p = 0.0011) 80 . Within the BR.21 trial, high EGFR gene copy or amplification also was associated with a significantly higher response to erlotinib (n = 91: 21% vs. 5%, p = 0.02) 71,77,85 . Similar findings were observed in the ISEL trial (n = 317: 16.4% vs. 3.2%) 83 .
In INTACT 1 and 2, there were no differences in response with and without EGFR amplification (n = 235: 56% vs. 53%, p > 0.05) 80 . Interestingly, analysis of tumour samples from the TRIBUTE study demonstrated a lower response rate among patients whose tumours demonstrated EGFR amplification 82,84 . It is important to note that FISH was used to assess EGFR gene copy status in the BR.21, ISEL, and TRIBUTE studies [82][83][84][85] , but that quantitative PCR was used in the IDEAL and INTACT studies 80 . High EGFR gene copy by FISH includes cases of EGFR high polysomy and of EGFR amplification alike [82][83][84][85] , but quantitative PCR results include cases of EGFR gene amplification only 80 . Thus, the two results are not entirely comparable.

Molecular Predictors of Survival:
The IDEAL 1 and 2 trials, BR.21, and ISEL all examined single-agent EGFR-TKIs 71,77,83,85 . Analysis of tumour samples from IDEAL 1 and 2 showed no significant improvement in TTP or survival for patients with EGFR mutations or with EGFR amplification 80 . However, these trials were not designed to examine predictors of survival, given that both groups of patients received an EGFR-TKI 80 .
The molecular analysis of the ISEL trial demonstrated a differential effect of gefitinib on survival according to EGFR gene copy (n = 370: FISH + HR 0.61 vs. FISH -HR 1.16; interaction p = 0.045) and EGFR expression (n = 379: IHC + HR: 0.77; IHC -HR: 1.57; interaction p = 0.049). The data were insufficient for a survival analysis for patients with and without EGFR mutations 83 .
Molecular analyses are available from all four trials evaluating the addition of an EGFR-TKI to platinumbased chemotherapy. The addition of gefitinib to 39 39 39 39 CURRENT ONCOLOGY-VOLUME 16, NUMBER 1    86 . Survival analysis from the TRIBUTE trial demonstrated a borderline improvement in TTP for patients receiving chemotherapy plus erlotinib (TTP HR: 0.59; 95% CI: 0.35 to 0.99), but no difference in OS for patients with EGFR amplification (n = 245) 82,84 . In patients with an EGFR mutation, there was also a trend toward improved TTP (12.5 months vs. 6.6 months, p = 0.092), but no difference in OS was demonstrated (p = 0.96, n = 274) 82,83 . Similar findings were observed in the TALENT study. The presence of EGFR mutations did not predict for improved OS (p = 0.65 for placebo vs. p = 0.40 for erlotinib) and PFS (p = 0.74 for placebo vs. p = 0.18 for erlotinib) irrespective of treatment 74,75 .
Information is more consistent that the presence of KRAS mutations is associated with worse survival for patients receiving an EGFR In contrast, there is no evidence that these molecular markers predict a differential effect on survival from an EGFR-TKI than from chemotherapy. The molecular analyses from the INTEREST trial showed no significant differences in survival between patients treated with gefitinib or with docetaxel according to EGFR expression, EGFR gene copy, EGFR mutational status, or KRAS status 79 .
Consensus Recommendation: Molecular markers such as EGFR high gene copy or EGFR mutations and clinical characteristics such as adenocarcinoma, female sex, never smoking, and Asian ethnicity appear to be associated with a higher likelihood of response to an EGFR-TKI. The evidence is currently insufficient to select patients based on molecular markers predictive of improved survival with an EGFR-TKI. Prospective data will be needed before further recommendations can be made.
The evidence is conflicting about the predictive value of clinical characteristics for survival. However, the data suggest that the survival benefit of an EGFR-TKI may be greater among never smokers. Based on available data, molecular markers and clinical characteristics should not be used to exclude patients from receiving EGFR-TKI therapy.

DISCUSSION
The EGFR-TKIs of represent a significant advance in the management of advanced and metastatic NSCLC. Not only do they have activity in NSCLC, they also appear to have an improved toxicity profile as compared with standard chemotherapy agents such as docetaxel. As a result, they offer an attractive therapeutic option. Nevertheless, it is important that these agents be incorporated into routine treatment algorithms based on appropriate data from randomized trials.
It is clear that EGFR-TKIs should not be used concomitantly with standard chemotherapy agents in the treatment of NSCLC. The strongest evidence supporting their use is in patients who have progressed following platinum-based chemotherapy. It is appealing to think that use of an EGFR-TKI may spare patients the toxicity of more chemotherapy. However, available data support the use of second-line chemotherapy and thirdline EGFR-TKI or second-line EGFR-TKI and then thirdline chemotherapy. Because both approaches prolong survival, the goal of therapy in advanced NSCLC should be to maximize the number of patients who receive three lines of therapy, if survival is the outcome of interest. However, some patients will choose not to have second-line chemotherapy, and so the sequence of therapies should reflect a discussion between the physician and the patient regarding the relative benefits and side effects of each treatment option.
Multiple reports in the literature suggest that molecular markers and clinical characteristics can be used to select patients who will be more likely to benefit from an EGFR-TKI. However, this literature comes with significant limitations. The term "benefit" creates confusion, because it is used to refer to a variety of outcomes, including tumour response, improved OS, and improved symptom control and QOL. The molecular analyses are limited to patients whose tumour samples were available. The percentage of patients whose samples were available for one or more molecular analyses ranged from 25% to 44% of the total study population. As a result, some of these comparisons involve small numbers of patients. In addition, much of the literature has focused on tumour response rates, rather than on survival. Although there is some consistency in factors predicting response, these factors do not correlate directly with variables predicting a differential benefit in survival. Considerable variation is found in the variables reported to be associated with a differential improvement in survival from therapy with an EGFR-TKI. This variation may exist in part because some of the EGFR markers are prognostic and associated with trends toward better survival (some EGFR mutations) or worse survival (high EGFR copy number). Therefore, it is not possible to assess the effect of EGFR-TKI therapy on survival in the absence of 44 44 44 44 44 a no-treatment control arm. Furthermore, markers that seem to predict for a differential survival benefit when EGFR-TKI therapy is compared with placebo or no treatment may not be predictive when EGFR-TKI therapy is compared with another form of treatment such as chemotherapy. As a result, the evidence is currently insufficient to recommend the routine use of molecular markers and clinical characteristics to select patients for therapy with an EGFR-TKI. It is therefore also premature to recommend the use of single-agent EGFR-TKIs as first-line therapy for NSCLC, even in patients selected on basis of molecular and clinical characteristics.
These results highlight the need for prospective trials in which tumour samples are available for all patients, so as to address correlative questions. Ongoing research will also address questions concerning the sequence of platinum-based chemotherapy or EGFR-TKI as first-line therapy.
Since the literature search for the present review was completed, preliminary data from two trials of maintenance gefitinib or erlotinib in Asian populations were presented at the American Society of Clinical Oncology Annual Scientific Meeting in 2008 87,88 . Both trials showed improved PFS, but no significant improvements in OS. In addition, initial results of IPASS (Iressa Pan ASia Study) were presented at the 2008 meeting of the European Society for Medical Oncology 89 . That trial compared first-line gefitinib with carboplatin and paclitaxel in light-or never-smoking Asian patients. A significant improvement was observed in PFS, but no significant difference in OS. Other ongoing trials are evaluating the role of an EGFR-TKI as maintenance therapy in patients responding to first-line platinumbased chemotherapy.
Lastly, chemotherapy experience suggests that the therapeutic ratio can be improved with combination therapy. Preliminary evidence suggests that combination therapy with an EGFR-TKI and agents active against vascular endothelial growth factor may have greater activity. These questions are being addressed in multiple ongoing clinical trials. Participation in these trials should be encouraged.

ACKNOWLEDGMENTS
This consensus statement was supported by an unrestricted educational grant from Hoffmann-La Roche Limited. The recommendations are those of the participants and are independent of Hoffmann-La Roche.