Chemo-radiotherapy versus surgery-based treatment for Stage IIIA non-small cell lung cancer: a systematic review and network meta-analysis

  • Protocol
  • Intervention

Authors


Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

To assess the effectiveness and safety of chemo-radiotherapy with surgery for people with Stage IIIA non-small cell lung cancer (NSCLC).

Background

Description of the condition

In 2012, there were 1.8 million new cases of lung cancer and 1.59 million deaths from the disease worldwide (International Agency for Research on Cancer 2012). Lung cancer is the most common cancer among men (1.2 million new cases), with the highest incidence rates in Central and Eastern Europe and Eastern Asia in 2012 (International Agency for Research on Cancer 2012). Among women, the incidence rates of lung cancer are lower (0.6 million new cases), and the highest incidence rates are in Northern America and Northern Europe in 2012 (International Agency for Research on Cancer 2012).

The World Health Organization classifies lung cancers into two major groups: non-small cell lung cancer (NSCLC) and small cell lung cancer (Travis 2015). More than 80% of all lung cancer cases are NSCLC. There are two major types of NSCLC: non-squamous carcinoma (e.g. adenocarcinoma, large cell carcinoma) and squamous cell carcinoma. Adenocarcinoma is the most common type of lung cancer.

Twenty per cent of all lung and bronchial cancers are diagnosed when the cancer has spread to regional lymph nodes or beyond the primary site (Howlader 2017). The majority of the lung cancer cases (57%) are diagnosed when the disease has metastasised (spread) to distant sites. The Tumor, Node, Metastasis (TNM) staging system, proposed by the American Joint Committee on Cancer (AJCC), is a widely accepted system to characterize the extent of NSCLC.The TNM staging staging system is based on the extent of tumour (T), extent of spread to the lymph nodes (N) and the presence of metastases (M).The numbers after each letter give more details about the cancer. For example, a higher number after the T indicates a larger tumor while a higher number after the N suggests more lymph nodes are involved by the cancer.The TNM system has undergone several changes over the last 40 years.

  1. The first edition of the AJCC TNM system, published in 1977, defined Stage III disease as T3 with any N or M, N2 with any T or M and M1 with any T or N (Beahrs 1977).

  2. The second edition, published in 1983, broadened the Stage III disease definition by including N1 disease with any T or M (Beahrs 1983).

  3. The third edition, published in 1988, first introduced the category of Stage IIIA disease which is defined by T1 N2, T2 N2, T3 N0-2 (Beahrs 1988).

  4. There was no difference in the definition of Stage IIIA disease between the fourth edition (published in 1992) (Beahrs 1992) and third edition (Beahrs 1988).

  5. The fifth edition, published in 1997, modified the category of Stage IIIA disease slightly by removing T3N0 disease (Fleming 1997).

  6. The definition of Stage IIIA disease remained the same between the sixth edition (published in 2002) (Greene 2002) and fifth edition (Fleming 1997).

  7. The seventh edition, published in 2010, broadened the definition of Stage IIIA disease slightly by including T4N0-1 disease (Edge 2010).

  8. The eighth edition, published in 2017, modified the Stage IIIA disease category even further to include only T1-2N2, T3N1, T4N0-1 (Amin 2017).

The prognostic significance of multi-station N2 disease was highlighted in the eighth edition of the AJCC TNM staging system, in which an exploratory subclassification of pathologic N2 disease was introduced. Presence of multi-station N2 disease is classified as pN2b disease whereas single N2 disease can be categorized as pN2a1 or N2a2 depending on the presence of concurrent N1 involvement. The overall survival rate for people with pN2b disease is significantly lower than for those with pN2a disease,and five-year overall survival rate of 38% (pN2b) versus 49% (pN2a) (Amin 2017).

Pearson and colleagues reported a series of 141 N2 patients who had undergone mediastinal lymph node examination and subsequent thoracotomy (an operation to allow the surgeon to access the chest cavity). They demonstrated that for patients who had a negative mediastinal lymph node evaluation, but had N2 disease identified at subsequent thoracotomy, there were higher rates of five-year overall survival compared with patients who had N2 disease at both mediastinal lymph node evaluation and subsequent thoracotomy (24% versus 9%) (Pearson 1982).

The location and definition of thoracic lymph nodes has changed over the years. Naruke and colleagues first proposed the thoracic lymph node map in the 1960s and it was widely adopted in the North America, Europe and Japan (Naruke 1967; Naruke 1978). The map was further refined by the American Thoracic Society (Tisi 1983), and by Mountain et al (Mountain 1997), and was subsequently adopted in the fourth to sixth editions of the TNM staging system. There were important differences between the maps by Naruke (Naruke 1967; Naruke 1978), and by Mountain (Mountain 1997). The level 7 subcarinal lymph nodes in the Mountain map corresponded to the levels 7 and 10 in the Naruke map. This resulted in some tumours being staged as N2 disease according to the Mountain map but as N1 disease according to Naruke map.

To reconcile the differences between the maps by Naruke (Naruke 1967; Naruke 1978), and Mountain (Mountain 1997), the members of the International Association for the Study of Lung Cancer (IASLC) developed a revised lymph node map which was adopted in the seventh (Edge 2010), and eighth (Amin 2017), editions of the TNM staging system. It is important to note that the definition of lymph node stations was changed in the seventh edition of the TNM staging system (Edge 2010). For example, the lymph node station 4R below the azygos vein — considered as N2 disease in the sixth edition of the TNM staging system (Greene 2002) — is now N1 disease (station 10) based on the seventh edition of the TNM staging system (Edge 2010).

Complete resection is a major prognostic factor for patients undergoing lung cancer surgery (Osarogiagbon 2012; Osarogiagbon 2013). Rami-Porta and colleagues have proposed clear definitions of what is considered a complete resection, incomplete resection or uncertain resection (Rami-Porta 2005). These definitions are consistent with the residual tumor classification , i.e. R0 for no identifiable tumour remaining, negative surgical margins; R1 for microscopic residual disease; and R2 for gross residual disease. Initially there was a subdivision named Run, indicating that a complete resection was probably achieved but not all criteria were fulfilled, which was proposed but not adopted. Examples of Run include incomplete nodal dissection when less than the required three hilar/intrapulmonary nodal stations, or less than three mediastinal nodal stations, were removed. Recently, the eighth edition of the TNM staging system has formally adopted the Run category (Amin 2017) (Appendix 1).

Description of the intervention

Surgery-based treatment approaches for Stage IIIA disease involve surgery with other treatment modalities such as chemotherapy and radiotherapy (National Comprehensive Cancer Network 2018). Surgical intervention entails anatomic pulmonary resection, such as lobectomy (excision of a lobe) or pneumonectomy (complete removal of a lung), as well as N1 and N2 lymph node resection and mapping with a minimum of three N2 stations sampled or complete lymph node dissection. In anatomically appropriate situations where margin-negative resection can be achieved, lung sparing anatomic resection (sleeve lobectomy) is preferred to pneumonectomy. Video assisted thoracic surgery (VATS) should be considered in patients with no anatomic contraindications as it has been shown to be associated with fewer complications and shorter time in hospital compared with lobectomy by thoracotomy (Cao 2013). The addition of chemotherapy, either preoperatively or postoperatively, has been shown to improve overall survival in patients with Stage IB to Stage IIIA NSCLC. An individual patient data meta-analysis, including 2385 patients, demonstrated that adding preoperative chemotherapy improved overall survival by five per cent at five years, from 40% to 45% in Stage IB-IIIA NSCLC (NSCLC Collaborative Group 2014). A Cochrane individual patient data meta-analysis, including 11,107 patients, reported that there was a clear absolute survival benefit of four per cent at five years from the addition of adjuvant chemotherapy, regardless of whether chemotherapy was given in addition to surgery or surgery plus radiotherapy for Stage IB to IIIA NSCLC (Burdett 2015).The role of adding radiotherapy to surgery for Stage IIIA NSCLC is unclear. A Cochrane individual patient data meta-analysis, including 2343 patients, showed that postoperative radiotherapy is detrimental to those with completely resected NSCLC and should not be used in the routine treatment of such patients (Burdett 2016). However, these trials employed now outdated radiotherapy techniques. Randomized trials are awaited to determine the effects of modern post-operative radiotherapy in patients with N2 disease. The role of adding radiotherapy to preoperative chemotherapy for NSCLC is unclear. A randomized trial comparing sequential neoadjuvant chemo-radiotherapy with neoadjuvant chemotherapy in Stage IIIA/N2 NSCLC showed that radiotherapy did not add any benefit to induction chemotherapy followed by surgery (Pless 2015). As this trial only compared preoperative sequential chemo-radiotherapy with preoperative chemotherapy, it is unclear if there will be a benefit with preoperative concurrent chemo-radiotherapy over chemotherapy alone for this group of patients.

Non-surgical treatment includes combined chemo-radiotherapy treatment. A randomized trial from 1968 involving 800 patients with locally advanced NSCLC, which compared radiotherapy alone and placebo, showed that the use of radiotherapy was associated with a small improvement in survival (18% at one year versus 14% in the control group) (Roswit 1968). A Cochrane review using individual patient data from 22 trials comparing radical radiotherapy with radical radiotherapy plus chemotherapy showed that there was a 10% reduction in the risk of death (corresponding to absolute benefits of 3% at two years and 2% at five years) with the use of radical radiotherapy plus chemotherapy (NSCLC Collaborative Group 2000). An individual patient data meta-analysis of six trials (1205 patients) showed that concurrent chemo-radiotherapy improved overall survival (4.5% absolute benefit at five years) compared with sequential chemo-radiotherapy in unresectable NSCLC (Auperin 2010).

How the intervention might work

Surgery, radiotherapy and chemotherapy are the three modalities used to treat patients with NSCLC according to the National Comprehensive Cancer Network (NCCN) guidelines (National Comprehensive Cancer Network 2018). They can be used alone or in combination depending on the stage of the disease, patients' performance status, comorbidities and pulmonary function. A patient can have resectable lung cancer by virtue of having a surgically removable NSCLC, but may not be operable due to poor pulmonary function or multiple comorbities.

For resectable and operable Stage I or II NSCLC, surgical resection should be used whenever possible. Postoperative chemotherapy can be given in pathologic Stage II NSCLC to reduce disease recurrence and improve survival. Patients with Stage I or II NSCLC who are inoperable or who refuse surgery may be candidates for definitive radiotherapy treatment, delivered using stereotactic or conventional techniques (National Comprehensive Cancer Network 2018).

For Stage III disease, a combined modality approach is used. When surgical resection is clearly feasible, surgery followed by postoperative chemotherapy with or without radiotherapy depending on the findings at surgery is an option. For N2 disease, options include preoperative chemotherapy or chemo-radiotherapy, possibly followed by surgery. When surgical resection is not feasible, the options include combinations of radiotherapy with sequential or preferentially concurrent chemotherapy (National Comprehensive Cancer Network 2018).

Why it is important to do this review

The optimal management of patients with resectable Stage IIIA NSCLC is unclear. The NCCN guidelines recommend a surgery-based approach for patients with resectable T1-3, N0-1 disease, and definitive concurrent chemo-radiotherapy for N2 disease (National Comprehensive Cancer Network 2018). The European Society for Medical Oncology (ESMO) guidelines also recommend surgery-based treatment for patients with N0-1 disease. However, for patients with potentially resectable N2 disease, the ESMO guidelines recognise that both surgery- and non-surgery-based treatment approaches are reasonable options (Eberhardt 2015).

Objectives

To assess the effectiveness and safety of chemo-radiotherapy with surgery for people with Stage IIIA non-small cell lung cancer (NSCLC).

Methods

Criteria for considering studies for this review

Types of studies

We will include randomized controlled trials that meet the eligibility criteria. The included trials must be published as full text. We will consider randomized trials regardless of language status. As these trials include surgical interventions, the blinding of participants or use of sham surgical procedure (i.e. placebo control) can be difficult and so they will not be mandatory requirements.

Types of participants

We will include adults (aged 18 years or older) with histologically diagnosed non-small cell lung cancer (NSCLC), staged by the 3rd to 8th editions of American Joint Committee on Cancer (AJCC) Tumor, Node, Metastasis (TNM) system.

Types of interventions

Intervention arm: chemo-radiotherapy. This arm will include radical chemo-radiotherapy treatment which can be given concurrently or sequentially. The minimum acceptable radiation dose in terms of equivalent dose in 2 Gray (Gy) per fraction (EQD2) assuming tumour alpha/beta ratio of 10 is 60 Gy. The use of targeted therapy or immunotherapy in this arm is allowed.

  1. Chemo-radiotherapy (can be concurrent or sequential)

  2. Chemo-radiotherapy + targeted therapy

  3. Chemo-radiotherapy + immunotherapy

Comparison arm: surgery-based treatment. This arm must include surgery with curative intent. Lung surgery will include lobectomy or pneumonectomy. Video assisted thoracic surgery is allowed. Adequate thoracic lymph node management should include N1 and N2 lymph node resection and mapping with a minimum of three N2 stations sampled or complete lymph node dissection. The use of chemotherapy, radiotherapy, targeted therapy and immunotherapy in this arm will be allowed.

Therefore, all the possible combinations of treatment are as follows.

  1. Surgery alone

  2. Surgery + chemotherapy (can be preoperative or postoperative, or both)

  3. Surgery + chemo-radiotherapy (can be preoperative or postoperative, or both)

  4. Surgery + targeted therapy +/- chemotherapy or chemo-radiotherapy

  5. Surgery + immunotherapy +/- chemotherapy or chemo-radiotherapy

We plan to conduct both traditional meta-analysis and network meta-analysis (NMA). In the meta-analysis, we will focus on broad intervention and comparison arms (i.e. group all the possible combinations into surgery arm and chemo-radiotherapy arm), while in NMA all the possible combinations will be considered as separate arms. All interventions will be compared to each other using a network meta-analysis (Figure 1). We assume that any participant that meets the inclusion criteria is, in principle, equally likely to be randomized to any of the eligible interventions.

Figure 1.

Network geometry of all interventions

Types of outcome measures

Primary outcomes

1. Overall survival (OS): survival until death from all causes. We will assess survival from the time of randomization. This will be reported as hazard ratio.

2. Adverse events (AEs): we will define the nature, timing and management of these events according toindividual trial protocols. We will examine the included trials for the incidence of Grade 5 adverse events, i.e. treatment related deaths. We will also look at incidence of Grade 3 or 4 pulmonary, cardiac and hematological adverse events (National Cancer Institute 2018). This will be reported as risk ratio.

Secondary outcomes

3. Disease-free survival (DFS): survival until recurrence of disease or death from all causes. We will assess disease-free survival from the time of randomization. This will be reported as hazard ratio.

4. Locoregional recurrence-free survival (LRFS): survival until locoregional disease recurrence or death from all causes. We will define locoregional disease recurrence according to individual trial protocols. We will assess locoregional recurrence-free survival from the time of randomization. This will be reported as hazard ratio.

5. Distant recurrence-free survival (DRFS): survival until distant disease recurrence or death from all causes. We will define distant disease recurrence according to individual trial protocols. We will assess distant recurrence-free survival from the time of randomization. This will be reported as hazard ratio.

6. Health-related quality of life (HRQL): the measurement of health-related quality of life outcomes, assessed using a scale that has been validated through reporting of norms in a peer-reviewed publication. An example of such a scale is the European Organization for Research and Treatment of Cancer (EORTC) quality of life questionnaire (Aaronson 1993). This will be reported as mean difference between treatment arms.

Search methods for identification of studies

We will adopt the search strategies recommended by the Cochrane Lung Cancer Group. The search strategies are based on the use of free-text terms and controlled vocabulary and are linked with the Cochrane highly sensitive search strategy for identifying randomized trials in MEDLINE, as referenced in Chapter 6.4.11.1 and detailed in box 6.4c of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Electronic searches

We will search the following electronic databases for eligible studies. We will not apply any language restrictions. If necessary, we will translate the non-English papers and assess them for potential inclusion in the review.

  1. Databases of medical literature

    1. Cochrane Central Register of Controlled Trials (CENTRAL; latest issue) (Appendix 2)

    2. MEDLINE, accessed via PubMed (1946 to present) (Appendix 3)

    3. Embase (1980 to present) (Appendix 4)

  2. Databases of ongoing trials

    1. the ISRCTN register of controlled trials:www.controlled-trials.com

    2. EU Clinical Trials Register: www.clinicaltrialsregister.eu/ctr-search/search

    3. US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov:www.clinicaltrials.gov

    4. WHO international trials registry platform: http://apps.who.int/trialsearch

Searching other resources

We will handsearch reference lists of included studies, relevant chapters of books, and review articles for eligible trials. We will also search the following, from 2015 onwards.

  1. The proceedings of the annual meetings of the American Society of Clinical Oncology (ASCO) and the American Society for Radiation Oncology (ASTRO)

  2. Lung cancer sections of the proceedings of the European Society of Medical Oncology (ESMO) Congress

  3. Lung cancer sections of the proceedings of the European Conference of Clinical Oncology (ECCO) Congress

  4. Proceedings of the World Conference on Lung Cancer

Data collection and analysis

We will summarize the evidence using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Selection of studies

Two review authors (YYS, WYK) will independently screen titles and abstracts of all identified studies, and exclude those which do not meet the inclusion criteria. We will retrieve the full-text copies of potentially relevant studies. Two review authors (YYS, WYK) will independently assess the eligibility of retrieved papers. We will identify and exclude duplicates and use the trial report with the most up-to-date results. We will contact the investigators when appropriate to clarify study eligibility. Disagreements will be resolved by discussion between the two review authors and if necessary by the third review author (IWKT). We will document the reasons for exclusions.

Data extraction and management

For included studies, two review authors (YYS, WYK) will independently extract the following study characteristics using a standardized data extraction form.

  1. Study methodology: study design, randomization, total duration of study, duration of follow-up period, numbers of centres and location and withdrawals

  2. Particpants: total sample size, numbers enrolled in each arm, mean age, age range, gender, pathological confirmation of NSCLC, NSCLC histological subtype, staging of NSCLC, staging system used, smoking history, inclusion and exclusion criteria.

  3. Interventions

    1. Chemo-radiotherapy arm: radiation dose fractionation schedule, types of radiation technique, types of chemotherapy use, timing of chemotherapy in relation to radiation, total number of cycles of chemotherapy, use of targeted therapy and immunotherapy in relation to radiotherapy

    2. Surgery-based treatment: types of lung surgery, types of thoracic lymph node resection, use of chemotherapy, radiotherapy, targeted therapy and immunotherapy in relation to surgery

  4. Outcomes

    1. For time-to-event data (survival and disease recurrence), we will extract the log of the hazard ratio (log HR) and its standard error from trial reports; if these are not reported, we will estimate the log (HR) and its standard error using published methods (Parmar 1998; Tierney 2007).

    2. For dichotomous data (adverse events), we will extract the number of patients in each treatment arm who experienced the outcome of interest and the number of patients assessed at the endpoint to estimate a risk ratio.

    3. For continuous data (health-related quality of life), we will extract the final value and standard deviation of the outcome of interest and the number of participants assessed in each treatment arm at the endpoint at the end of follow-up to estimate the mean difference between treatment arms and its standard error.

We will extract both adjusted and unadjusted statistics if reported. We will extract data that are relevant to an intention-to-

treat analysis in which participants were analysed in the groups to which they were assigned. We will also note the time points at which outcomes were collected.

Assessment of risk of bias in included studies

Two review authors (YYS, QZ) will evaluate the risk of bias for each study independently using the criteria recommended in the Cochrane 'Risk of bias' tool (Higgins 2011). Disagreements will be resolved by discussion between the two review authors and if necessary by the third review author (IWKT). We will assess the risk of bias according to the following domains.

  1. Random sequence generation (per study)

  2. Allocation concealment (per study)

  3. Blinding of participants and personnel (per study)

  4. Blinding of outcome assessment (per outcome)

  5. Incomplete outcome data (per outcome)

  6. Selective outcome reporting (per outcome)

  7. Other bias.

We will classify each domain as high, low or uncertain risk of bias and provide a quote from the study report together with a justification for our judgement in the 'Risk of bias' table. We will summarize the 'Risk of bias' judgements across different studies for each listed domain.

Measures of treatment effect

We will analyse time-to-event data (survival, disease recurrence) using hazard ratios if possible. We will analyse dichotomous outcomes (adverse events) using risk ratios if possible. We will analyse continuous outcomes (health-related quality of life) using mean differences between treatment arms if possible.

We will undertake meta-analyses only when the treatments, participants and the underlying clinical question are similar enough for pooling to be appropriate. When multiple trial arms are reported in a unique study, we will include only the relevant arms. If two comparisons for continuous or dichotomous outcomes must be included into the same meta-analysis, we will halve the control group to avoid double counting.

Unit of analysis issues

We will analyse each eligible trial for potential unit of analysis errors, such as using non-standard trial design or reporting multiple observations for the same outcomes. We believe that these errors will be uncommon in this systematic review. If a trial reports multiple observations for the same outcomes, we will choose the observation with the most frequently reported time point across the trials, and conduct sensitivity analyses based on which time point is chosen.

Dealing with missing data

We will contact the trial investigators to obtain missing numerical outcome data where possible. We will not impute missing outcome data for the primary outcome.

Assessment of heterogeneity

We will use the I2 statistic to assess heterogeneity among the trials in each pairwise comparison, for both standard meta-analysis and direct comparisons within network meta-analysis (Higgins 2011). If there is evidence of substantial heterogeneity (I2 greater than 50%) in the standard meta-analysis, we will explore and report the possible causes using prespecified subgroup analysis.

We will assess heterogeneity of both meta-analysis and direct comparisons within the network meta-analysis, as network meta-analysis needs to fulfil three assumptions, i.e. heterogeneity (for direct comparisons), similarity (for indirect comparisons) and consistency (for combining direct and indirect comparisons) (Donegan 2013).

Assessment of similarity (transitivity)

We will assess the assumption of transitivity by comparing the population, intervention, comparison, and outcomes across different comparisons to identify potential sources of effect modifiers.

Assessment of loop inconsistency

We will adopt the side-splitting method to evaluate the potential loop inconsistency within each comparison between estimates from direct and indirect evidence. This method evaluates the consistency assumption in each closed loop of the network separately as the difference between direct and indirect estimates for a specific comparison in the loop (Dias 2010). Then, the magnitude of the inconsistency factors and their 95% confidence intervals (CIs) can be used to infer about the presence of inconsistency in each loop.

Assessment of design inconsistency

To check the potential design inconsistency, we will use the "design-by-treatment" model (Higgins 2012; White 2012). This method accounts for different sources of inconsistency that can occur when studies with different designs (two-arm trials versus three-arm trials) give different results, as well as disagreement between direct and indirect evidence (i.e. loop inconsistency). Using this approach, we will infer about the presence of inconsistency from any source in the entire network based on a global Wald test.

Assessment of reporting biases

We will assess selective reporting within each trial by comparing the methods reported in the protocol (if available) and final published study. We will try to contact the trial investigators if reporting bias is suspected. If this is not possible and the missing data are thought to introduce serious bias, we will explore the impact of including such studies in the overall assessment of results by conducting a sensitivity analysis.

If there are ten or more trials available for pooling, we will investigate publication bias using a funnel plot.

Data synthesis

We will first perform the standard meta-analysis comparing chemo-radiotherapy arm with surgery based treatment arm using a random-effects model (Deeks 2011). For time-to-event data (i.e. overall survival, disease-free survival, locoregional recurrence-free survival and distant recurrence-free survival), we will use the inverse variance method. We will use the Mantel-Haenszel method for handling binary outcomes (adverse events).

We will perform network meta-analysis using a random-effects model in STATA with the mvmeta command within the network suite of commands for network meta-analysis (White 2015), and other STATA commands for visualising and reporting results in network meta-analysis (Chaimani 2015). If meta-analysis is not possible, we will undertake a narrative review of the findings.

We will present a 'Summary of findings' table to identify the key results of the review (Higgins 2011). We will include the following outcomes: overall survival, adverse events, disease-free survival, locoregional disease-free survival, distant disease-free survival and health-related quality of life. We will also demonstrate the illustrative risk of these outcomes and define the absolute and relative magnitude of effect.

We will use the five GRADE criteria (study limitations, consistency of effect, imprecision, indirectness and publication bias) to evaluate the quality of a body of evidence as it relates to the studies which contribute data to the meta-analysis for each prespecified outcome. We will use GRADEpro software (GRADEpro GDT 2015). We will justify all decisions to downgrade or upgrade the quality of evidence using footnotes and we will make comments to facilitate the readers' comprehension.

Subgroup analysis and investigation of heterogeneity

We will consider the following subgroup analyses where possible.

  1. Types of staging system: AJCC third/fourth edition versus fifth/sixth edition versus seventh edition versus eighth edition

  2. Types of Stage IIIA: N2 versus N1

  3. Types of lung surgery: pneumonectomy versus lobectomy

  4. Adequacy of lymph node dissection: adequate versus inadequate or uncertain adequacy

  5. Resection status: complete resection only versus incomplete resection or uncertain resection status

  6. Types of radiation technique: intensity modulated radiation therapy versus non-intensity modulated radiation therapy

  7. Timing of chemotherapy in relation to radiation: concurrent versus sequential

  8. Definition of survival time point: before versus after induction therapy

Sensitivity analysis

We will perform sensitivity analyses by excluding studies with a high risk of bias. Adequate random sequence generation and allocation concealment are important for ensuring the study quality. Hence, we will categorize those studies which are at high risk of bias for either of these domains as having high risk of bias overall, and will exclude them for this sensitivity analysis.

Acknowledgements

We acknowledge the help and support of the Cochrane Lung Cancer Group and particularly Paul Van Schil, Mia Schmidt-Hansen, John Ruckdeschel, Fergus Macbeth, Corynne Marchal and Virginie Westeel, as well as the Information Specialists Francois Calais and Giorgio Maria Agazzi for designing our search strategies.

Appendices

Appendix 1. T,N,M descriptors for the eighth edition of TNM classification for lung cancer

T: Primary tumour
TXPrimary tumour cannot be assessed or tumor proven by presence of malignant cells in sputum or bronchial washings but not visualized by imaging or bronchoscopy
T0No evidence of primary tumour
TisCarcinoma in situ
T1Tumour <=3cm in greatest dimension surrounded by lung or visceral pleura without bronchoscopic evidence of invasion more proximal than the lobar bronchus (i.e. not in the main bronchus)
T1a(mi)Minimally invasive adenocarcinoma
T1aTumour <= 1cm in greatest dimension
T1bTumour > 1cm but <= 2cm in greatest dimension
T1cTumour > 2cm but <= 3cm in greatest dimension
T2

Tumour > 3cm but <= 5cm or tumour with any of the following features:

-involves main bronchus regardless of distance from the carina but without involvement of the carina

-invades visceral pleura

-associated with atelectasis or obstructive pneumonitis that extends to the hilar regionm involving part or all of the lung

T2aTumour < 3cm but <= 4cm in greatest dimension
T2bTumour > 4cm but <= 5cm in greatest dimension
T3Tumour > 5cm but <= 7cm in greatest dimension or associated with separate tumour nodule(s) in the same lobe as the primary tunour or directly invades any of the following structures: chest wall (including parietal pleural and superior sulcus tumours), phrenic nerve, parietal pericardium
T4Tumour > 7cm in greatest dimension or associated with separate tunour nodule(s) in a different ipsilateral lobe than that of the primary tumour or invades any of the following structures: diaphragm, mediastinum, heart, great vessels, trachea, recurrent laryngeal nerve, esophagus, vertebral body, and carina
N: Regional lymph node involvement
NXRegional lymph nodes cannot be assessed
N0No regional lymph node metastasis
N1`Metastasis in ipsilateral peribronchial and/or ipsilateral hilar lymph nodes and intrapulmonary nodes, including involvement by direct extension
N2Metastasis in ipsilateral mediastinal and/or subcarinal lymoh node(s)
N3Metastasis in contralateral mediastinal, contralateral hilar, ipsilateral or contralateral scalene, or supraclavicular lymph node(s)
M: Distant metastasis 
M0No distant metastasis
M1Distant metastasis present
M1aSeparate tumour nodule(s) in a contralateral lobe; tumour with pleural or pericardial nodule(s) or malignant pleural or pericardial effusion
M1bSingle extrathoracic metastases
M1cMultiple extrathoracic metastases in one or more organs
Stage groupings
Occult carcinomaTXN0M0
Stage 0TisN0M0
Stage IA1

T1a(mi)N0M0

T1aN0M0

Stage IA2T1bN0M0
Stage IA3T1cN0M0
Stage IBT2aN0M0
Stage IIAT2bN0M0
Stage IIB

T1a-2bN1M0

T3N0M0

Stage IIIA

T1a-2bN2M0

T3N1M0

T4N0-1M0

Stage IIIB

T1a-2bN3M0

T3-4N2M0

Stage IIICT3-4N3M0
Stage IVAAny T, Any N, M1a-b
Stage IVBAny T, Any N, M1c

Appendix 2. CENTRAL search strategy

#1MeSH descriptor: [Carcinoma, Non-Small-Cell Lung] explode all trees
#2nsclc
#3lung cancer*
#4lung carcinom*
#5lung neoplasm*
#6lung tumor*
#7lung tumour*
#8non small cell*
#9nonsmall cell*
#10(#3 or #4 or #5 or #6 or #7) and (#8 or #9)
#11#1 or #2 or #10
#12MeSH descriptor: [Thoracic Surgical Procedures] explode all trees
#13surger*
#14surgic*
#15resect*
#16MeSH descriptor: [Pneumonectomy] explode all trees
#17pneumonectom*
#18MeSH descriptor: [Thoracotomy] explode all trees
#19thoracotom*
#20preoperat*
#21postoperat*
#22lobectom*
#23#12 or #13 or #14 or #15 or #16 or #17 or #18 or #19 or #20 or #21 or #22
#24MeSH descriptor: [Radiotherapy] explode all trees
#25MeSH descriptor: [Chemoradiotherapy] explode all trees
#26radiother*
#27chemorad*
#28chemo-radiother*
#29chemother*
#30radiation therap*
#31MeSH descriptor: [Drug Therapy] explode all trees
#32antineoplas*
#33MeSH descriptor: [Antineoplastic Agents] explode all trees
#34#24 or #25 or #26 or #27 or #28 or #29 or #30 or #31 or #32 or #33
#35advanced
#36IIIa
#373a
#38#35 or #36 or #37
#39#23 and #34 and #38
#40#11 and #39

Appendix 3. MEDLINE search strategy

"#54,""Search #42 AND #53""
"#53,""Search #51 NOT #52""
"#52,""Search animals [MeSH Terms] NOT humans [MeSH Terms]""
"#51,""Search #43 OR #44 OR #45 OR #46 OR #47 OR #48 OR #49 OR #50""
"#50,""Search groups[Title/Abstract]""
"#49,""Search trial[Title/Abstract]""
"#48,""Search randomly""
"#47,""Search drug therapy[MeSH Subheading]""
"#46,""Search placebo[Title/Abstract]""
"#45,""Search randomized[Title/Abstract]"
"#44,""Search controlled clinical trial[Publication Type]""
"#43,""Search randomized controlled trial[Publication Type]""
"#42,""Search #11 AND #41""
"#41,""Search #23 AND #36 AND #40""
"#40,""Search #37 OR #38 OR #39""
"#39,""Search 3a""
"#38,""Search IIIa""
"#37,""Search advanced""
"#36,""Search #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32 OR #33 OR #34 OR #35""
"#35,""Search antineoplastic agents[MeSH Terms]""
"#34,""Search antineoplas*""
"#33,""Search drug therapy[MeSH Terms]""
"#32,""Search radiation therap*""
"#31,""Search chemother*""
"#30,""Search chemo-radiother*""
"#29,""Search chemorad*""
"#28,""Search radiother*""
"#27,""Search chemoradiotherapy[MeSH Terms]""
"#26,""Search Radiotherapy[MeSH Terms]""
"#23,""Search #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22""
"#22,""Search lobectom*""
"#21,""Search postoperat*""
"#20,""Search preoperat*""
"#19,""Search Thoracotom*""
"#18,""Search thoracotomy[MeSH Terms]""
"#17,""Search Pneumonectom*""
"#16,""Search pneumonectomy[MeSH Terms]""
"#15,""Search resect*""
"#14,""Search surgic*""
"#13,""Search surger*""
"#12,""Search thoracic surgical procedures[MeSH Terms]""
"#11,""Search #1 OR #2 OR #10""
"#10,""Search (#3 OR #4 OR #5 OR #6 OR #7) AND (#8 OR #9)""
"#9,""Search nonsmall cell*[Title/Abstract]""
"#8,""Search non small cell*[Title/Abstract]""
"#7,""Search lung tumour*[Title/Abstract]""
"#6,""Search lung tumor*[Title/Abstract]""
"#5,""Search lung neoplasm*[Title/Abstract]""
"#4,""Search lung carcinom*[Title/Abstract]""
"#3,""Search lung cancer*[Title/Abstract]""
"#2,""Search nsclc[Title/Abstract]""
"#1,""Search Carcinoma, Non-Small-Cell Lung[MeSH Terms]""

Appendix 4. Embase search strategy

#42 #11 AND #23 AND #36 AND #40 AND #41
#41 'crossover procedure'/exp OR 'double-blind procedure'/exp OR 'randomized controlled trial'/exp OR 'single-blind procedure'/exp OR random* OR factorial* OR crossover* OR cross NEXT/1 over* OR placebo* OR doubl* NEAR/1 blind* OR singl* NEAR/1 blind* OR assign* OR allocat* OR volunteer*
#40 #37 OR #38 OR #39
#39 3a:ti,ab
#38 iiia:ti,ab
#37 advanced:ti,ab
#36 #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32 OR #33 OR #34 OR #35
#35 'molecularly targeted therapy'/exp
#34 'immunotherapy'/exp
#33 'antineoplastic agent'/exp
#32 antineoplas*:ti,ab
#31 'drug therapy'/exp
#30 'radiation therap*':ti,ab
#29 'chemother*':ti,ab
#28 'chemo rad*':ti,ab
#27 chemorad*:ti,ab
#26 radiother*:ti,ab
#25 'chemoradiotherapy'/exp
#24 'radiotherapy'/exp
#23 #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22
#22 lobectom*:ti,ab
#21 postoperat*:ti,ab
#20 preoperat*:ti,ab
#19 thoracotom*:ti,ab
#18 'thoracotomy'/exp
#17 pneumonectom*:ti,ab
#16 'lung resection'/exp
#15 resect*:ti,ab
#14 surgic*:ti,ab
#13 surger*:ti,ab
#12 'thorax surgery'/exp
#11 #1 OR #2 OR #10
#10 #3 OR #4 OR #5 OR #6 OR #7 AND (#8 OR #9)
#9 'nonsmall cell*':ti,ab
#8 'non small cell*':ti,ab
#7 'lung tumour*':ti,ab
#6 'lung tumor*':ti,ab
#5 'lung neoplasm*':ti,ab
#4 'lung carcinom*':ti,ab
#3 'lung cancer*':ti,ab
#2 'nsclc':ti,ab
#1 'non small cell lung cancer'/exp

Contributions of authors

Protocol development: YYS, QZ WYK, CNL, LS, EC, IWKT

Selection of trials for inclusion: YYS, QZ WYK, CNL IWKT

Methodological assessment and data extraction: YYS, QZ, WYK, CNL, LS, EC IWKT

Data entry and management: YYS, QZ

Data analysis: YYS, QZ, LM, EC

Preparation of manuscript: YYS, QZ, WYK, CNL, LS, EC IWKT

Declarations of interest

Yu Yang Soon: none known

Qishi Zheng: none known

Luming Shi: none known

Edwin SY Chan: none known

Cheng Nang Leong: none known

Wee Yao Koh: none known

Ivan Weng Keong Tham received sponsorship-in-kind from Elekta KK, honorarium from Celgene Pte Ltd for speaking services in 2016 and honorarium for ESMO Preceptorship Programme for Lung Cancer in 2017

Ancillary