Epidermal Growth Factor Receptor (EGFR) Tyrosine Kinase Inhibitors versus single agent chemotherapy as second‐line treatment for non–small‐cell lung cancer wild‐type or unknown status for EGFR
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
The aim of this systematic review is to evaluate the effectiveness and safety of EGFR‐TKIs versus single‐agent chemotherapy as second‐line treatments in patients with advanced NSCLC and wild‐type or unknown status for EGFR.
Background
Description of the condition
Lung cancer remains the leading cause of cancer‐related death worldwide causing approximately 1.6 million deaths per year with a five‐year survival less than 15% (Ferlay 2015). Non–small‐cell lung cancer (NSCLC) represents approximately 85% of lung cancers. Most of the patients are diagnosed with advanced‐stage disease unsuitable for surgery or radical radiotherapy. After first‐line platinum‐based doublet chemotherapy, a second‐line therapy is recommended for patients with performance status (PS) from 0 to 2 who have progressed clinically or radiologically. Mono‐chemotherapy remains the standard treatment in this setting by American Society of Clinical Oncology (ASCO) (Masters 2015) and European Society of Medical Oncology (ESMO) (Novello 2016) Clinical Practice Guidelines.
In the last decade, genetic alterations have been identified in NSCLC tumours, corresponding to key oncogenic events called driver oncogenic alterations. The first driver oncogenic alteration identified in this setting were in the epidermal growth factor receptor (EGFR) and more recently anaplastic lymphoma kinase (ALK) gene rearrangement. However, EGFR mutations and ALK gene rearrangement are found in only a minority of patients (11% and 5% of Caucasian patients, respectively) (Barlesi 2016). Even in Asia‐Pacific populations, with the highest EGFR mutation frequency (47% for non‐squamous cell carcinoma (Midha 2015)), a significant proportion of patients do not have tumours with EGFR mutations and are so called 'wild‐type' EGFR. Therefore, worldwide, the majority of patients with NSCLC have tumours that are wild‐type for EGFR (Laurie 2013).
The use of routine mutation testing varies within and between countries. In many places, routine EGFR mutation testing is not carried out (Midha 2015), in which case, the proportion of patients with unknown status for EGFR increases. Sometimes, even when EGFR mutation testing is available, it is not technically feasible for some patients if, for instance, there is insufficient tissue.
Finally, despite the current commitment to 'personalized medicine' and the increased availability of DNA sequence testing in NSCLC, the majority of randomized trials up to now, have been conducted in an unselected population.
Description of the intervention
Cytotoxic chemotherapy was the first treatment assessed in second‐line NSCLC treatment in a randomized controlled trial (RCT) comparing docetaxel to best supportive care performed in 2000 (Shepherd 2000). Since then, docetaxel has been considered as the reference treatment in this setting. Cytotoxic chemotherapy is composed of a variety of drugs with different mechanisms of action, all aiming at stopping cell division and consequently tumour growth. For example, docetaxel is an antineoplastic agent that disrupts the microtubular network in cells. This leads to the stabilization of microtubules, which results in the inhibition of mitosis in cells and so stops cell division. A second cytotoxic drug (pemetrexed) was approved for second‐line treatment of advanced NSCLC, and secondarily restricted to non‐squamous cell carcinoma. Pemetrexed is a folate analog metabolic inhibitor that disrupts folate‐dependent metabolic processes essential for cell replication. Sometimes, older molecules are still used in this indication, such as gemcitabine, vinorelbine, and paclitaxel.
In the last decade, targeted treatments have emerged with the first of these targeting the EGFR signalling pathway. The EGFR family of genes, which is frequently expressed in solid tumours, encodes a widely expressed transmembrane glycoprotein pertaining to the ErbB (Erythroblastic Leukemia Viral Oncogene Homolog) receptor tyrosine kinase superfamily. Tyrosine Kinase Inhibitors (TKIs), binding to the intracellular domain of the tyrosine kinase, inhibit EGFR downstream signalling and block the EGFR‐mediated cancer cell proliferation and propagation. Whereas EGFR‐TKIs were more effective in advanced NSCLC harbouring EGFR mutations, in all stages of NSCLC patients without EGFR mutations treated with EGFR‐TKI, the response rate reached 11% and the disease control rate 42% (Lindeman 2013). Erlotinib was the first EGFR‐TKI approved in second‐line treatment of advanced NSCLC. Gefitinib, another EGFR‐TKI, was also recognized as a second‐line treatment by the American Society of Clinical Oncology (ASCO) Clinical Practice Guideline (Masters 2015).
In this review, we will compare EGFR‐TKIs to cytotoxic single‐agent chemotherapy.
Why it is important to do this review
Six overlapping systematic reviews and meta‐analyses (Qi 2012; Gao 2013; Lee 2014; Li 2014; Vale 2014; Zhao 2014), published from October 2012 to November 2014, assessed EGFR‐TKIs versus chemotherapy. However, these reviews had several important limitations. Primary studies included in these systematic reviews varied; the reviews included between three and 14 trials, with a median of seven trials. Despite similar inclusion criteria, only two trials were included in all of them. Only one systematic review searched clinicaltrials.gov; and none searched industry trial registries and results databases, or regulatory agency online databases. Results of time to progression were included in one meta‐analysis (Li 2014), instead of those of progression‐free survival.
One systematic review considered overall population (Qi 2012), four focused on patients with wild‐type EGFR status (Gao 2013; Lee 2014; Vale 2014; Zhao 2014) and one considered patients according to their EGFR mutation status (Li 2014). Nevertheless, the majority of trials evaluating EGFR‐TKIs versus chemotherapy were performed in an unselected population without consideration of patients’ EGFR mutation status at randomisation. Therefore, for the majority of trials, retrospective subgroup analyses based on a posteriori analysis of EGFR mutation status, have been performed. The previous systematic reviews and meta‐analyses considered only the results of these subgroups of patients identified retrospectively. However, these post‐hoc analyses, not pre‐specified with incomplete sample acquisition (considering only a subset of the randomised patients, usually less than one‐third), are not adequate evidence to support biomarker results and introduce bias in pooled results in the meta‐analysis. Besides, the consistency and the reliability of the diagnosis of truly EGFR wild‐type patients in each trial is critical, because of the variability of detection methods for EGFR mutations (e.g. Polymerase Chain Reaction (PCR) ‐based highly sensitive methods versus direct sequencing).
Considering all the points previously mentioned, it seems relevant to carry out a systematic review and meta‐analysis to compare EGFR‐TKIs with single‐agent chemotherapy as second‐line treatment in patients with advanced NSCLC, considering patients wild‐type or unknown status for EGFR.
Objectives
The aim of this systematic review is to evaluate the effectiveness and safety of EGFR‐TKIs versus single‐agent chemotherapy as second‐line treatments in patients with advanced NSCLC and wild‐type or unknown status for EGFR.
Methods
Criteria for considering studies for this review
Types of studies
We will consider only RCTs.
Types of participants
We will consider trials that included participants aged more than 18 years old with an advanced (stage III unsuitable for radical radiotherapy or surgery and stage IV) NSCLC receiving second‐line systemic treatment. We will exclude trials that included only patients harbouring EGFR‐activating mutation or ALK translocation.
Types of interventions
We will consider RCTs comparing any EGFR‐TKI as a single agent (e.g. erlotinib or gefitinib) to a single‐agent cytotoxic chemotherapy including docetaxel and pemetrexed. We will not consider trials assessing a combination of two cytotoxic drugs (because in the Di Maio meta‐analysis of individual patient data, which analysed six trials (847 participants), doublet chemotherapy increased response rate and progression‐free survival but was more toxic and did not improve overall survival compared to single‐agent therapy (Di Maio 2009)).
We will consider RCTS of second‐line therapy. However, we will also consider trials including both second‐ and third‐line therapy, because in these specific trials, patients with third‐line therapy are a minority and are clinically similar to patients with second‐line therapy.
We will include trials in which patients in the control arm received different single‐agent chemotherapy agents at the investigators’ discretion (e.g. docetaxel or pemetrexed). We will exclude trials concerning maintenance therapy, but patients can have received prior maintenance therapy before second‐line treatment.
Types of outcome measures
The following outcome measures will be considered in this review.
Primary outcomes
-
Overall Survival (OS), defined as the time from randomisation to death from any cause.
-
Toxicity measured by the number of participants with serious adverse events (AE) as defined atclinicaltrials.gov (clinicaltrials.gov/ct2/about‐studies/glossary#S), and with severe adverse events (treatment‐related grade 3 or above toxicity in Common Terminology Criteria for Adverse Events (CTCAE 2009)).
Secondary outcomes
-
Progression‐Free Survival (PFS), defined as the time from randomisation to disease progression or death from any cause.
-
Objective response, defined as a complete response or a partial response according to the Response Evaluation Criteria in Solid Tumors (RECIST) (Therasse 2000).
-
Quality of life (QoL), according to the standardised scale used in the included trials.
Search methods for identification of studies
Electronic searches
We will search the following databases from inception to present:
-
Cochrane Lung Cancer Group Specialised Register;
-
Cochrane Central Register of Controlled Trials (CENTRAL) (Cochrane Library);
-
MEDLINE, accessed via PubMed;
-
Embase.
There will be no restriction on the language, type, or year of publication.
We will search all databases using both controlled vocabulary (namely MeSH in MEDLINE and EMTREE in Embase) and a wide range of free‐text terms. The search of MEDLINE will be performed using the Cochrane highly sensitive search strategy and precision‐maximising version (2008 version) as described in the Cochrane Handbook for Systematic Reviews of Interventions (Chapter 6.4.11.1 and detailed in box 6.4.b)) (Higgins 2011a) .
The search equations for MEDLINE (accessed via PubMed) and CENTRAL are reported in Appendix 1.
The search equation for Embase is reported in Appendix 2.
Searching other resources
-
We will search previous systematic reviews in the Cochrane Library, the Database of Abstracts of Reviews of Effects and the PROSPERO international prospective register of systematic reviews for completed or published systematic reviews.
-
We will screen the reference lists of all trials.
-
We will screen the proceedings of the following conferences:
-
-
ASCO from 2013;
-
ESMO from 2013;
-
World Conference on Lung Cancer (WCLC) from 2013.
-
-
We will search in non‐industry trial registries and results databases (clinicaltrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform) (www.who.int/ictrp) to identify ongoing and unpublished trials.
-
We will search industry trial registries and results databases of each identified company to search for ongoing and unpublished trials.
-
We will search reviews submitted to the US Food and Drug Administration and to the European Medicines Agency for drug registration on regulatory agency online databases.
-
For trials identified as 'completed' in clinicaltrials.gov, but without posted results or those identified only by a conference proceeding, we will send trialists a personalised email to request complete results, with a reminder at a later date.
Data collection and analysis
Selection of studies
Two authors (PC, AY) will independently and in duplicate examine each title and abstract identified in the search to exclude obviously irrelevant reports. The two authors will then independently examine full‐text articles to determine eligibility. We will contact trial authors for clarification, when necessary. We will discuss any disagreements with a third author (LT) to reach consensus. We will list studies and document the primary reasons for exclusion.
We will perform the whole study selection process using a web service developed at the French Cochrane Centre (http://cochrane‐resyweb.net).
Data extraction and management
Two authors (PC, AY) will independently extract the data from published and unpublished reports using a standardised form. We will discuss disagreements with a third author (LT) to reach consensus. We will check the data and enter them into the Cochrane Review Manager computer software (RevMan 2014). We will contact authors of trials to provide missing outcome data.
We will extract from each included trial:
-
trial characteristics: study phase, single‐centre or multicentre status, funding source (private, public, both, or unclear), number of randomised patients
-
description of treatment: drugs, dosage, frequency and modality of administration
-
patients characteristics: age, gender, stage (IIIB versus IV), histology (non‐squamous versus squamous cell carcinoma), ethnicity (Asian versus Caucasian), performance status PS (PS 0‐1 versus PS 2), history of smoking (never versus former or current smoker), EGFR mutation status (wild‐type versus unknown EGFR status), KRAS mutation status, proportion of patients in second‐line treatment.
-
outcome data: hazard ratios (HRs) for PFS and OS and their 95% confidence intervals (CIs), number of patients with an objective response (sum of partial response and complete response), number of patients with serious and severe AE, and means and standard deviations for QoL.
In case of several reports pertaining to the same trial, we will extract outcome data results from the different sources giving priority to the first available source among: regulatory agency reports, results posted on clinicaltrials.gov, full‐text articles, pharmaceutical reports, and conference abstracts.
Assessment of risk of bias in included studies
Two review authors (PC, AY) will assess the risk of bias using the 'risk of bias' tool described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). We will discuss disagreements with a third author (LT) to reach consensus. We will assess the risk of bias by using the six evidence‐based domains: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), and selective outcome reporting (reporting bias). For each risk of bias domain, the authors will assign either 'low risk of bias', 'high risk of bias' or 'unclear risk of bias'.
The blinding domain will be assessed at the outcome level, regarding blinding for objective outcomes (OS) and for subjective outcomes (PFS, objective response, serious and severe AE, and quality of life).
-
For blinding of participants and care providers, studies will be considered at 'low risk' if blinding was assured or if the outcome was unlikely to be influenced by the lack of blinding (OS); and studies will be considered at 'high risk' for subjective outcomes if blinding was lacking.
-
For blinding of outcomes assessors, studies will be considered at 'low risk' if blinding was assured or for an objective outcome. If an independent Clinical Endpoint Adjudication Committee assessed subjective outcomes, studies will be considered at 'low risk' and at 'high risk' otherwise.
We will complete a 'risk of bias' table for each included study and we will also summarize the risks of bias across studies.
Measures of treatment effect
For each outcome, we will calculate summary estimates of treatment effects (with 95% CI) for each comparison. For time‐to‐event endpoints (OS, PFS), we will report HRs. When time‐to‐event data are unavailable from trial reports, we will reconstruct individual survival patient data from published Kaplan‐Meier curves and we will estimate hazard ratios (Guyot 2012).
For dichotomous outcomes (objective response and toxicity), we will use odds ratios (OR). Where sufficient data are reported, ORs will be calculated for each toxicity category from the total number of grade three and four events in each treatment arm.
For continuous outcomes (QoL), mean differences (MD) will be presented for measures using the same scale, and standardized mean differences (SMD) will be presented for measures that used different scales.
Unit of analysis issues
The primary unit of analysis will be the participant. We are not expecting cross‐over or cluster‐randomised trials.
Dealing with missing data
Where data are missing or unsuitable for analysis, we will contact the authors using their email addresses from the trial reports, from trial registers or from the authors' institutions to request further information. However, if we do not retrieve the missing data: for survival outcomes, we will exclude these trials, and for binary data, we will perform an available case analysis and we will run different sensitivity analyses, making different assumptions for the missing participants.
Assessment of heterogeneity
The studies will be evaluated clinically and methodologically to assess if it is reasonable to consider combining data. We will assess statistical heterogeneity by a visual inspection of the forest plots and statistically through an assessment of homogeneity based on the Chi2 test, for which a P value of less than 0.10 will be considered an indication of substantial heterogeneity. The I² statistic, which indicates the percentage of the variability in effect estimates because of true between‐study variance rather than sampling error (within‐study variance), will be calculated (Higgins 2002). The criterion for identification of substantial heterogeneity will be an I² statistic value of greater than 50%. If significant heterogeneity is identified, we will determine the source of the observed heterogeneity (any outlying studies driving this heterogeneity) and we will explore its potential causes.
Assessment of reporting biases
To address reporting bias and related small study effects, we will construct funnel plots for each pairwise meta‐analysis, although interpretation is difficult if there is only a small number of trials (< 10). If the required statistical conditions are met (≥ 10 studies, no significant heterogeneity, and ratio of the maximal to minimal variance across studies > 4), we will use asymmetry tests (Ioannidis 2007; Rücker 2008).
Data synthesis
We will synthesise the data using Review Manager 5.3 (RevMan 2014). We will undertake meta‐analyses only if data are judged to be sufficiently similar to ensure an answer that is clinically meaningful. We will consider the results from fixed‐effect and random‐effects models. The choice between the two models will be based on the number of trials, the distribution of effect sizes, the assessment of heterogeneity and of clinical and methodological diversity. For OS and PFS, the estimated log HR and variance of individual trials will be combined. The combined HR represents the overall risk of an event in the experimental group versus the control group. Absolute differences in median OS and PFS will be estimated using the HR and the average control group median OS or PFS ((median/HR) ‐ median). Where data aggregation is not possible, the results of individual studies will be presented in tables or graphics and discussed. Similar methods will be used for dichotomous and continuous outcomes.
Summary of findings table
We will create a 'summary of findings' table using the following outcomes: OS; PFS; objective response rate; serious AE; QoL. We will use the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of a body of evidence as it relates to the studies which contribute data to the meta‐analyses for the prespecified outcomes (Guyatt 2008). We will use methods and recommendations described in section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). We will justify all decisions to downgrade or upgrade the quality of studies using footnotes and we will make comments to aid the reader's understanding of the review, where necessary.
Subgroup analysis and investigation of heterogeneity
If sufficient studies are available, we will perform subgroup analyses:
-
EGFR status (trials performed in patients with wild‐type EGFR versus trials performed in an unselected population for EGFR status);
-
Histology (non‐squamous cell carcinoma versus a mix of non‐squamous and squamous cell carcinoma);
-
Ethnicity (Asian versus Caucasian versus mixed population);
-
History of smoking (trials specifically performed in never smokers versus the others).
Sensitivity analysis
If relevant, we will conduct sensitivity analyses to assess if the results are robust to decisions made during the review process (for example, regarding eligibility criteria, by excluding trials considering patients receiving both second‐ and third‐line treatment).
