Criteria for considering studies for this review
Types of studies
Randomised controlled trials and quasi-randomised controlled trials comparing radiotherapy and surgery as part of treatment for localised Ewing's sarcoma in children.
Types of participants
The review will include trials on children (less than 18 years of age) with a diagnosis of localised Ewing’s sarcoma established by histopathology and a negative metastatic workup. There will be no restriction by participant gender or ethnic group. We will also consider studies which include participants younger than 18 years and older than 18 years. For such studies we will assess if separate data are available for participants younger than 18 years. If data are not given separately for our review population we will not include these studies in the review.
Types of interventions
We will include trials which compare radiotherapy versus surgery for localised Ewing’s sarcoma in addition to neo-adjuvant and adjuvant chemotherapy. The type and dose of chemotherapy must be the same in both intervention and control groups.
Types of outcome measures
Event-free survival (EFS) is defined as the time from the date of diagnosis until the event or last participant contact, at which time the participant is censored. An 'event' will be defined as stated by the authors of included studies. Overall survival is defined as the time from the date of diagnosis until the death of the participant from any cause.
Risk of second malignancy, defined as cancer that develops after treatment for a first cancer. The risk of growth retardation will be determined in terms of limb length discrepancy.
We will investigate the above-mentioned outcomes, although these outcome measures are not used as criteria for inclusion of studies. Studies will be chosen purely on their relevance to the study objective.
Search methods for identification of studies
We will not impose any language or date of publication restriction. We will update searches every two years.
We will search the following electronic databases for primary studies: the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, latest issue), MEDLINE/PubMed (from 1945 to present), EMBASE/Ovid (from 1980 to present). The Cochrane Childhood Cancer Group (CCG) will run the searches in these databases.
The search strategies for the different electronic databases (using a combination of controlled vocabulary and text words) are shown in the appendices (Appendix 1; Appendix 2; Appendix 3).
Searching other resources
We will locate information about trials not identified in CENTRAL, MEDLINE and EMBASE, either published or unpublished, by searching the reference lists of relevant trial articles and review articles.
We also plan to scan the conference proceedings of the International Society of Paediatric Oncology (SIOP) (from 2009 to present), the European Musculo-Skeletal Oncology Society (E.M.S.O.S.) from 2012 onwards, the International Society of Limb Salvage (ISOLS) (2011 and 2013) and the American Society of Clinical Oncology (ASCO; meetinglibrary.asco.org/abstracts) (from 2009 to present).
We will search for ongoing trials by scanning the following trial registers:
We will also handsearch the journals in which the included studies most frequently appear. We will search the top five journals (according to the number of included studies provided) for the last 12 months.
We will contact authors of relevant papers regarding any further published or unpublished work.
Data collection and analysis
Selection of studies
Two review authors (IQ and JD) will download all titles and abstracts retrieved by electronic searching to the reference management database Endnote, will remove duplicates and examine the remaining references independently. These two review authors will screen titles and abstracts of references identified from the search and will eliminate articles that are obviously not relevant to the search question. We will assess all other references in full text. Each review author will then independently determine if these trials are eligible for inclusion, resolving disagreements about inclusion by discussion and if necessary by recourse to a third review author (MU).We will contact study authors for further details when papers contain insufficient information to make a decision about eligibility. We will document reasons for exclusion. We will include a flow chart/diagram of the search in the review.
Data extraction and management
Two review authors (IQ and NA, not masked to the study details) will independently perform extraction of data from the included trials, using a data extraction form designed and pilot-tested. We will extract each trial component, such as study design, participants, setting, intervention and control, co-interventions, outcomes and results, follow-up duration, funding source details, the declaration of interests for the primary investigators, and risk of bias domains, and will present them in a table of Characteristics of Included Studies.
The review authors will compare results and resolve disagreements by discussion until a consensus is reached. If there is no consensus, a third review author (MU) will settle the discrepancies. We will document relevant disagreements. We will write to the study authors for any missing, incomplete or unclear data. One review author (IQ) will enter the data in to Review Manager 5 software (RevMan 2012) and another (JD) will check the data entered manually. A third author (ZL) will check the whole process.
Dealing with duplicate publications
If there are multiple publications for one study, we will use the one with the most recent data. We may use earlier publications for extracting information on baseline characteristics or methodology.
Assessment of risk of bias in included studies
Two review authors (IQ and JD) will independently assess the risk of bias for each study (that is, selection bias, performance bias, detection bias (for each outcome separately), attrition bias (for each outcome separately), reporting bias and other potential bias) using the 'Risk of bias' items as described in the module of the Childhood Cancer Group (Module CCG), which are based on the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We will contact study authors if relevant data are missing. We will take the risk of bias in included studies into account in the interpretation of the review's results.
Measures of treatment effect
Measures of treatment effect will depend on the types of data presented in the individual studies.
1) Hazard ratios (HRs) for time-to-event variables (event-free survival and overall survival). We will use Parmar's method if hazard ratios have not been explicitly presented in the study (Parmar 1998).
2) Risk ratios (RRs) for dichotomous variables (risk of second malignancy and growth arrest).
Unit of analysis issues
Studies with more than two treatment groups
If we identify studies with more than two intervention groups (multi-arm studies), we will where possible combine groups to create a single pair-wise comparison or use the methods set out in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) to avoid double-counting study participants. For the subgroup analyses, when the control group is shared by two or more study arms, we will divide the control group (events and total population) over the number of relevant subgroups to avoid double-counting the participants.
Dealing with missing data
As stated before, when relevant data regarding study selection, data extraction and risk of bias assessment are missing, we will attempt to contact the study authors to retrieve the missing data. If after 30 days we cannot obtain this information, we will use the available information and perform an ‘available case analysis’ instead of an intention-to-treat analysis if possible. An ‘available case analysis’ includes data on only those participants whose results are known, using as a denominator the total number of people who had data recorded for the particular outcome in question (the participants are analysed according to the group to which they were randomised). If we do not have enough information to include a certain outcome in the analysis, we will declare this.
We will describe missing outcomes of the included studies by reporting proportions of randomised participants for whom no outcome data were obtained (with reasons) by outcome and by randomised group. We will address the potential impact of the missing outcomes on the results of the included studies in the assessment of risk of bias and we will describe in the Discussion section its impact on the findings of the review. We will perform sensitivity analyses to assess how sensitive results are to changes in the assumptions made in the ‘available case analysis.’
Assessment of heterogeneity
We will investigate meta-analysis results for clinical and statistical heterogeneity.
We will check for heterogeneity by considering the following factors:
1. The clinical or methodological characteristics of the studies.
2. The results of the I² statistic for the quantification of the heterogeneity. The I² statistic describes the percentage of total variation across studies that is due to heterogeneity rather than to chance (Higgins 2003). We will judge the importance of the observed value of I² depending on the magnitude and direction of effects and the strength of evidence for heterogeneity (moderate to high heterogeneity will be defined as I² of 50% or more).
If we find substantial heterogeneity, we will not pool the data, but will attempt to explain the observed differences by examining individual study characteristics and by performing subgroup analyses.
Assessment of reporting biases
In addition to the evaluation of reporting bias as described in the 'Assessment of risk of bias' section, we will investigate reporting biases (such as publication bias) using funnel plots, provided there are 10 or more studies in the meta-analysis. We will assess funnel plot asymmetry visually, and use formal tests for funnel plot asymmetry. For continuous outcomes we will use the test proposed by Egger 1997, and for dichotomous outcomes we will use the test proposed by Harbord 2006. If we detect asymmetry in any of these tests or if a visual assessment suggests it, we will perform exploratory analyses to investigate it.
We will carry out statistical analysis using the Review Manager 5 software (RevMan 2012). We will use a fixed-effect meta analysis for combining data where it is reasonable to assume that studies are estimating the same underlying treatment effect, i.e. where trials are examining the same intervention, and we judge the outcome definitions, study population and methods to be sufficiently similar between studies. If there is clinical heterogeneity (regarding interventions, study populations, methods and outcome definitions) we will not pool the results. We will descriptively summarise studies for which pooling of results is not possible. We will use the fixed-effect model if I² is 50% or less. If I² is greater than 50% we will use a random-effects model. We will present the results as the average treatment effect with 95% confidence intervals, and the estimate of I². We will set out the main findings of the review in a 'Summary of findings' table prepared using the GRADE approach using GRADE profiler software (Guyatt 2008). We will list all outcomes for each comparison with estimates of relative effects along with the number of participants and studies contributing data for those outcomes. For each individual outcome, we will assess the quality of the evidence using the GRADE approach, which involves consideration of within-study risk of bias (methodological quality), directness of evidence, heterogeneity, precision of effect estimates and risk of publication bias. We will rate the quality of the body of evidence for each key outcomes as 'high', 'moderate', 'low' or 'very low'.
Subgroup analysis and investigation of heterogeneity
We will undertake subgroup analysis, grouping the trials by radiation dose (< 60 Gy versus ≥ 60 Gy). The usual radiation dose in Ewing’s sarcoma in 40 to 60 Gy. Higher radiation doses are associated with increased risk of adverse events, including second malignancies, and lower doses are associated with suboptimal response (Kuttesch 1996).
We will assess subgroup differences by interaction tests available within RevMan. We will report the results of subgroup analyses quoting the Chi² statistic and P value, and the interaction test I² value.
Besides the sensitivity analyses mentioned earlier in this protocol, we will conduct a repeat of the primary meta-analysis, excluding studies at high risk or uncertain risk of bias. In this sensitivity analysis, we will only include studies that have a low risk of bias in all key domains for the estimates of treatment effect.