Criteria for considering studies for this review
Types of studies
We will include randomized controlled trials (RCTs) of any intervention for the management of brain radionecrosis in adult patients previously treated with radiation therapy to the brain. Because we anticipate a limited number of RCTs, we will also include all comparative prospective intervention trials and quasi trials of any intervention for brain radionecrosis in adult patients. We will include nonrandomized studies in the discussion and will consider them for inclusion in the analysis if we identify an acceptable comparison cohort, adjusted for major prognostic factors.
Types of participants
We will include studies that evaluate adult (> 18 years old) patients previously treated with radiosurgery or fractionated radiotherapy for brain tumor with a diagnosis of brain radionecrosis based on clinical and radiological criteria, with or without pathological confirmation, as tissue confirmation is not frequently acquired in clinical practice.
We will define a clinical and radiological diagnosis of brain radionecrosis as "a growing enhancing lesion with associated edema in the region of prior high-dose radiation in the presence of low suspicion of active tumor due to a long disease control interval, lack of tumor involvement within the brain (e.g., head and neck cancer), or advanced imaging evidence to suggest absence of active tumor.
As the purpose of this review is to provide evidence to guide clinical practice, and as pathological confirmation is often unavailable for patients who may have a combination of radionecrosis and tumor, this review will exclude patients who lack pathological confirmation in the evaluation of any treatment, including treatments that may be active against radionecrosis and tumor.
Types of interventions
For pharmacological interventions, we will investigate the efficacy and effectiveness of any dose of agent given by any route for the purpose of treating brain radionecrosis. Agents likely to be included are antioxidant agents such as vitamin E and antiangiogenic agents such as bevacizumab. To improve the clinical relevance of this review, we will compare the efficacy and effectiveness of these agents against standard clinical care, which typically includes corticosteroid therapy.
For nonpharmacological interventions, we will include any treatment given with the aim of treating brain radionecrosis to improve symptoms and prevent progression of the process. These treatments will likely include surgery and hyperbaric oxygen therapy.
Because of the limited number of RCTs anticipated, we will include all prospective intervention trials, including single-arm studies. For these studies, we will report efficacy and effectiveness and will consider these data against outcomes reported for standard care, which typically includes corticosteroid therapy.
Types of outcome measures
The primary outcome will be radiological response, defined as any reduction in contrast-enhancing lesions or edema (i.e., T2-weighted hyperintensity on MRI or hypodensity on computed tomography (CT)).
Clinical improvement, defined as documented physician-reported or patient-reported improvement in neurological status, symptoms, or functional independent status.
Corticosteroid requirements, reported as the ability of patients to decrease their corticosteroid dose or to stop corticosteroids completely.
Treatment-related adverse events, including death, hemorrhage, hematological toxicity, pulmonary toxicity, cardiac toxicity, gastrointestinal (GI) toxicity, and infection.
Quality of life, using scales such as the M.D. Anderson Symptom Inventory Brain Tumor Module (MDASI-BT) and the European Organization for Research and Treatment of Cancer core quality of life questionnaire (EORTC QLQ-C30).
Search methods for identification of studies
We will search the following electronic databases: the Cochrane Central Register of Controlled Trials (CENTRAL, current issue), MEDLINE (1950 to date), EMBASE (1980 to date), and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) (1982 to date). We have listed the MEDLINE search strategy in Appendix 1.
For databases other than MEDLINE, we will adapt the search strategy as needed. We will identify all relevant articles on PubMed, and we will use the "Related articles" feature to carry out a further search for newly published articles. We will apply no language restrictions in our searches.
Searching other resources
Unpublished and grey literature
We will search Metaregister, Physicians Data Query, www.controlled-trials.com/rct, www.clinicaltrials.gov, and www.cancer.gov/clinicaltrials to search for ongoing trials.
If through these searches we identify ongoing trials that have not been published, we will contact the principal investigators to request relevant data. We will likewise approach the major co-operative trial groups active in this area. We will search conference proceedings and abstracts through ZETOC (http://zetoc.mimas.ac.uk). We will search theses and dissertations through WorldCat (http://firstsearch.oclc.org).
We will handsearch the reference lists of included studies, key textbooks, and previous systematic reviews. We will handsearch journals and conference materials from the past year in the following sources.
Annual Meeting of the European Association of Neuro-Oncology (EANO).
Annual Congress of the European Society for Radiotherapy and Oncology (ESTRO).
Annual Meeting of the World Federation of Neuro-Oncology (WFNO).
Annual Meeting of the American Society of Clinical Oncology (ASCO).
Annual Meeting of the American Society for Therapeutic Radiation Oncology (ASTRO).
Annual Meeting of the American Society of Neuro-Oncology (SNO).
Bienniel Congress of the International Stereotactic Radiosurgery Society (ISRS).
Biennial Meeting of the Leksell Gamma Knife Society (LGKS).
Annual Meeting of the Multinational Association of Supportive Care in Cancer (MASCC).
We will undertake personal communication with authors of relevant trials and experts at major hospitals performing clinical trials to identify further data that may or may not have been published. We will seek papers in all languages and will carry out translations if necessary.
Data collection and analysis
Selection of studies
We will download all titles and abstracts retrieved by electronic searching to the reference management database, Endnote. We will remove duplicate references, and two review authors (CC, PB) will independently examine the remaining references. The review authors will not be blinded to study authors or to affiliations of the studies. We will exclude studies that clearly do not meet the inclusion criteria and will obtain copies of the full text of potentially relevant references. Two review authors (CC, PB) will independently assess the eligibility of retrieved papers. Review authors will resolve disagreements by discussion and will document reasons for exclusion.
Data extraction and management
For included trials, data will be abstracted as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Two review authors (CC, PB) will abstract data independently onto a data abstraction form specially designed for the review.
This form will include the following data.
Article details (author, year of publication, journal citation, country, and language).
Intervention (characteristics and duration).
Study design and methodology (including inclusion and exclusion criteria, assignment process, and timing of measurements).
Population demographics and total number involved; details of the health status of participants, including tumor histology, prior treatment details, and performance status.
Outcome measures (radiological response, clinical improvement, corticosteroid requirements, quality of life, and adverse events).
Risk of bias.
Data will be collated and entered into RevMan 5 2014.
When possible, all data extracted will be those relevant to an intention-to-treat (ITT) analysis, in which participants are analyzed in the groups to which they are assigned. The time points at which outcomes were collected and reported will be noted.
For continuous outcomes, we will record the final value and standard deviation of the outcome of interest and the number of participants in each treatment arm assessed at the end of follow-up, and we will use these values to estimate the mean difference between treatment arms and its standard error. We will note time points at which outcomes were collected and reported. We will resolve differences between review authors by discussion. For dichotomous variables, we will record the outcome of interest and the number of participants in each treatment arm assessed for an early time point (within four months of treatment) and at last follow-up.
Assessment of risk of bias in included studies
We will assess the risk of bias in included studies by using the tool of The Cochrane Collaboration (Higgins 2011a). This will include assessment of:
selection bias: random sequence generation and allocation concealment;
performance bias: blinding of participants and personnel (patients and treatment providers);
detection bias: blinding of outcome assessment;
attrition bias: incomplete outcome data;
reporting bias: selective reporting of outcomes; and
other possible sources of bias.
We will interpret all results of meta-analyses in light of findings with respect to risk of bias. We will judge all bias criteria and will report them as having low, high, or unclear risk of bias. We will classify the risk of bias as unclear when insufficient information is provided, or when there is uncertainty over the potential for bias. Two review authors (CC, PB) will apply the risk of bias tool independently and will resolve differences by discussion. We will summarize the results in both a risk of bias graph and a risk of bias summary. We will interpret results of meta-analyses in light of findings with respect to risk of bias.
Measures of treatment effect
For dichotomous outcomes (e.g., reduction in lesion volume or not, clinical improvement or not), we will record for each study the number of participants who experience the outcome of interest following treatment at an early time point (within four months of treatment) and at last follow-up.
For continuous outcomes (e.g., lesion volume, quality of life measures), we will express treatment effect as standardized mean differences between treatment arms with 95% confidence intervals, only if appropriate, and if a mean difference method is not possible.
Unit of analysis issues
Two review authors (CC, PB) will review unit of analysis issues according to information provided in the Cochrane Handbook for Systematic Reviews of Interventions and will resolve differences by discussion (Higgins 2011). These may include reports that describe:
individuals receiving more than one intervention (e.g., cross-over trial, simultaneous treatment of methods for each individual); or
multiple observations for the same outcome (e.g., repeated measurements, recurring events).
Dealing with missing data
We will not impute missing outcome data for any outcome. If data for the primary outcome are missing, we will contact trial authors to request data on outcomes among participants who were assessed.
We will include details of missing data in the narrative summary and "Risk of bias" table, alongside an assessment of the extent to which missing data could have altered the results of the review.
Assessment of heterogeneity
We will assess heterogeneity between studies by visual inspection of forest plots (including the presence of outliers and poor overlap of confidence intervals) and by a formal statistical test of the significance of the heterogeneity (Deeks 2001). When suggested by the forest plot and when I2 is greater than 50%, suggesting substantial heterogeneity (Higgins 2011; Higgins 2011a), we will further explore the causes of heterogeneity.
Assessment of reporting biases
Two review authors (CC, PB) will review and record reporting biases. If suitable, we will examine funnel plots to assess the potential for small-study effects such as publication bias.
We will pool data in a meta-analysis using RevMan 5 2014 when studies are comparable with respect to participants, interventions, and outcomes.
We will use the random-effects models with inverse variance weighting for all meta-analyses, along with a 95% confidence interval (DerSimonian 1986).
Subgroup analysis and investigation of heterogeneity
As we anticipate a limited number of eligible studies for this review, subgroup analyses may not be feasible but will be attempted for the following subgroups.
Radiosurgery versus fractionated radiotherapy.
Pathologically confirmed radionecrosis versus clinical-radiological diagnosis alone.
Tumor histology for which the initial radiotherapy was provided: brain metastases, primary central nervous system (CNS) tumors, other (likely head and neck cancer).
Whether sensitivity analysis will be required will be determined by agreement among all review authors and in keeping with information provided in Higgins 2011. We will consider the following factors as possible sources of heterogeneity across studies: variation in study quality (high or low levels of risk of bias), use of different classes of agents, and dosing or scheduling differences. We anticipate that additional possible types of sensitivity analyses will be identified during the conduct of the review.