Plain language summary
Surgical sampling or removal of low-grade glioma brain tumors
The issue: Low-grade gliomas (LGGs) are slow growing, less aggressive brain tumors. The most optimal surgical management is under debate.
The aim of the review: There are two surgical management strategies (treatments) for a person with a suspected LGG. These are biopsy, the surgical sampling of a small amount of tumor tissue, or resection, where as much as possible of the tumor is surgically removed. Tissues from both operations are then histologically examined to give a definitive diagnosis of the type and grade (severity) of the tumor. The aim of the review is to determine if biopsy or resection is the best treatment strategy.
The main findings: There is much debate about which of these surgical techniques is the best for patient survival. We searched the literature up to June 2016. However, currently there are no randomized controlled trials (RCTs) which have looked at which is the better procedure, only retrospective research studies looking at how people have responded to procedures that have happened in the past. Therefore, in the future, more RCTs are needed to try and answer this question.
Quality of the evidence: We were unable to determine this as no studies were included and only low-quality evidence from non-RCTs is available.
Conclusions: There are no randomized clinical trials on this topic; some institutional, non-clinical trials studies have suggested improved overall survival and seizure control with higher extent of resection. However, physicians should approach each case individually and weigh the risks and benefits of biopsy versus surgical resection, as well as incorporate patient preference into their clinical decision-making. Prognostic factors such as patient age, tumor size, and tumor location as well as potential implications for quality of life should be taken into account.
Prélèvement chirurgical d'échantillon ou ablation des tumeurs du cerveau (gliomes) de bas grade
Les gliomes de bas grade (GBG) sont des tumeurs du cerveau à croissance lente et peu agressives.
Deux stratégies de prise en charge chirurgicale sont possibles pour le traitement d'une personne chez qui on suspecte un GBG. Ce sont la biopsie, le prélèvement chirurgical d'une petite quantité de tissu tumoral, et la résection, où l'on enlève chirurgicalement le plus possible de la tumeur. Dans les deux cas, les tissus sont ensuite examinés pour établir un diagnostic définitif quant à la nature et au grade (gravité) de la tumeur.
La question de laquelle de ces procédures chirurgicales est préférable pour la survie du patient reste très débattue. Il n'y a toutefois pas actuellement d'essais cliniques randomisés ayant examiné quelle est la meilleure procédure, mais seulement des études rétrospectives portant sur la façon dont les gens ont répondu à des procédures réalisées dans le passé. C'est pourquoi de nouveaux essais de recherche randomisés seront nécessaires pour tenter de répondre à cette question.
Notes de traduction
Translated by: French Cochrane Centre
Translation supported by: Pour la France : Minist�re de la Sant�. Pour le Canada : Instituts de recherche en sant� du Canada, minist�re de la Sant� du Qu�bec, Fonds de recherche de Qu�bec-Sant� et Institut national d'excellence en sant� et en services sociaux.
This review is an update of a previously published review in the Cochrane Database of Systematic Reviews (2013, Issue 4)(Veeravagu 2013).
Description of the condition
Gliomas are a group of central nervous system (CNS) neoplasms consisting of neuroglial cells. Low-grade gliomas (LGGs) are rare and constitute approximately one fifth of all CNS glial tumors, and affect 1800 to 3000 new patients annually in the USA (Pouratian 2010). The incidence of low-grade astrocytoma has not been shown to vary significantly with nationality and has an incidence of 0.8 cases per 100,000 population (Epidemiology of LGG). Compared to high-grade gliomas (HGGs), LGG are heterogeneous, slower growing, less aggressive lesions, with patients typically living with the disease for five to 20 years. Efforts have been underway to consolidate our current understanding of the clinical behavior for these tumors in order to optimize management and to provide a basis for future randomized clinical trials.
Establishing a diagnosis of LGG is important in order to differentiate a lesion from a more aggressive tumor type. The differential diagnosis should also include non-neoplastic lesions, which must be ruled out. A presumptive diagnosis of LGG can be made based on clinical presentation and imaging characteristics. A patient with transient neurological symptoms consistent with seizure and a non-enhancing hemispheric mass lesion on magnetic resonance imaging (MRI) or computed tomography (CT), or both, that produces little mass effect are suggestive of LGG. Focal neurological deficits are rare but can occur. These tumors are best seen on T2-weighted and fluid attenuated inversion recovery (FLAIR) MRI sequences and are frequently non-enhancing on T1-gadolinium sequences. Currently it is unclear what the diagnostic accuracy of MRI is for suspected LGG. Clinically, studies have shown that among 1028 patients with brain tumor, the prevalence of seizures was higher in patients with LGGs (85%) than in patients with anaplastic glioma (69%) or glioblastoma (49%) (Lote 1998). Furthermore, in a series of 831 consecutive patients, tumor contrast enhancement was present on CT in 21% of cases with LGG compared with 57% to 96% of those with HGGs (Lote 1998b).
Although suspected LGG can be diagnosed clinically and radiographically, a definitive diagnosis must be made based on two histopathologic characteristics of a surgical specimen. First, the cell type that constitutes the bulk of the tumor is of glial origin, which can be further subdivided as astrocytoma, oligodendroglioma, or mixed oligoastrocytoma. Second, the grade of the tumor is rated based on the World Health Organization (WHO) classification scheme (WHO Tumor Grading Classification). For the purpose of this review, we restricted our definition of LGG to hemispheric WHO Grade II astrocytomas, oligodendroglioma, or mixed oligoastrocytoma. The controversy regarding biopsy versus resection lies specifically in the management of these diffuse hemispheric LGGs. A large proportion of LGGs, and virtually all diffuse LGGs, fall in the above mentioned three histologies.
There is general consensus that Grade I tumors such as pilocytic astrocytomas, subependymal giant cell astrocytoma, pleomorphic xanthoastrocytoma, subependymoma, etc. are managed with surgical resection, which is preferred as a cure; while other therapies are less effective. The diagnosis for WHO I tumors can frequently be suspected on imaging and surgical resection is then procured, thus the controversy regarding biopsy versus resection does not exist in these types of tumors.
In summary, LGGs constitute a class of slow-growing primary brain neoplasms. Optimal clinical management has been debated, with general agreement that a combination of surgical, radiation therapeutic, and chemotherapeutic approaches are necessary for optimal survival and outcome. The goals of treatment for patients with LGGs include prolonging overall survival (OS) and progression-free survival (PFS) and minimizing morbidity.
Description of the intervention
Management of LGGs differs from that of higher-grade lesions. If a LGG is suspected based on clinical presentation and radiographical characteristics, the most definitive diagnosis will require a histopathological and molecular diagnosis by obtaining a tissue sample (via either biopsy or resection).
There are many treatment dilemmas and one major discussion involves the role of surgery. Surgical biopsy includes all procedures that aim to obtain enough diagnostic tissue to make a definitive pathological diagnosis. The procedure may be performed open or with a needle with a freehand, stereotactic, or image-guided technique. Surgical resection includes all procedures where the pre-operative aim is to remove more tumor than is necessary for making a pathological diagnosis.
There remains controversy about the role of biopsy versus resection in the initial management of patients with LGG. The main aim of surgical resection is to improve survival, or at least delay the need for subsequent therapy. Resection may also improve neurological deficits and seizure control. However, gross total resection is often not possible without a significant risk of neurological sequelae due to the diffuse infiltrative nature of LGG. Biopsy can confirm diagnosis and carries fewer risks, but may not extend survival or improve symptoms, and may have an associated sampling error. Retrospective studies indicate that a maximally safe resection at the time of diagnosis may be linked with improved survival (Keles 2001; Pignatti 2002; Shaw 2002). In contrast, there is also evidence that a more conservative approach (delaying resection therapy until radiographic evidence of tumor growth, transformation, or impairment) may be more appropriate for patients with small, minimally symptomatic LGG (Olson 2000; Recht 1992; van Veelen 1998). More recently, the European Federation of Neurological Societies - European Association for Neuro-Oncology (EFNS-EANO) Task Force released guidelines which suggest that younger age, normal neurological examination, oligodendroglial histology, and chromosome 1p loss are favorable prognostic factors. Total or near total resection can improve seizure control, PFS and OS, all while reducing the risk of malignant transformation (Soffietti 2010).
How the intervention might work
Surgical resection in the management of LGG decreases tumor burden and may decrease the rate of progression or recurrence, or both, of the tumor. Biopsy of LGG provides a pathological or histological diagnosis but does not achieve debulking of the tumor. Usually resection involves an open procedure in which a craniotomy must be performed, whereas a biopsy may be performed through the use of a stereotactic needle.
Why it is important to do this review
The current treatment approach for patients with LGG includes operative management (biopsy or resection) followed by either immediate or delayed postoperative radiation therapy. The relatively long survival times of patients with LGG have made secondary outcomes including quality of life (QoL) measures and cognitive performance a significant component of management decisions. Thus, we have undertaken a systematic review of clinical trials that address optimal long-term outcomes in patients with LGG managed by surgical biopsy versus resection.
Specific to this study, it was hypothesized that resection may provide a clinical advantage over biopsy, which only provides histological confirmation. However, the current literature is conflicting on the relative merits of each procedure and it is not readily apparent whether the more invasive procedure of resection confers any long-term benefits in OS or PFS. Fairly unique among those with CNS malignancy, patients with LGG have long expected survival times. Thus, it is crucial to develop therapeutic approaches that optimize patient survival as well as cognitive performance and quality of life. The authors performed a systematic review of the literature on this topic to inform future clinical decisions on the initial surgical management of patients with LGG.
The literature surrounding low-grade gliomas (LGGs) is far more limited than that for high-grade gliomas (HGGs), possibly due to the fact that the major cause of mortality of LGGs is the advancement to HGG. Fortunately, the literature continues to expand and comprehensive reviews have recently been published on the treatment options for LGGs (Omay 2012; Ruiz 2009).
Our thorough examination of the literature via this Cochrane review did not identify any studies that met pre-defined inclusion criteria. At present, there remains a significant need for randomized controlled trials (RCTs) evaluating resection versus biopsy for LGGs. This area of study may benefit patients who maintain poor preoperative morbidity and are thus more likely to undergo radiotherapy rather than surgical resection. Such an RCT has been performed to evaluate resection versus biopsy in patients diagnosed with HGGs (Vuorinen 2003).
Pathological diagnosis via biopsy or resection is important for LGGs as malignancy may be underestimated using radiographic and clinical characteristics alone (Lote 1998). If a patient presents with a lesion that appears to be an LGG based on radiographic and clinical evidence, it remains unclear in the clinical trials literature whether definitive diagnosis via biopsy or resection will result in increased progression-free survival (PFS) and quality of life (QoL). However, it is well known that the larger sample obtained from surgical resection presents greater opportunity for revealing heterogeneity (Revesz 1993).
Furthermore, several recent institutional studies have demonstrated the advantages of maximal resection for patient outcomes. In a prospective, volumetric analysis of extent of resection (EOR) of 68 consecutive patients with hemispheric LGGs, smaller preoperative tumor volume and greater EOR were found to be associated with longer overall survival (OS), PFS and malignant degeneration-free survival (MFS) (Majchrzak 2012). In another prospective study on 86 patients with LGGs, both univariate and multivariate analysis demonstrated a statistical correlation between gross total removal and longer PFS (Jung 2011). Overall, Sanai and Berger reviewed every major glioma publication from 1990 to 2008, which included 10 LGG articles, and concluded that more extensive surgical resection is associated with longer life expectancy (Sanai 2008).
Finally, a recent study from two Norwegian universities examined survival in population-based parallel cohorts involving 153 patients with diffuse LGGs; 66 patients were enrolled at hospital A (47 received biopsy, 19 received initial resection) while 87 were enrolled at hospital B (12 biopsy, 75 resection). OS was better with early resection, at a rate of 74% at five years compared to 60% in the biopsy group (Jakola 2012). Regional practice variation played an important role in the type of treatment offered; at hospital A, resection of suspected LGG was only offered if a safe total resection was feasible based on preoperative planning, while at Hospital B, due to availability of neuro-navigation with three dimensional (3D) ultrasound-based intra-operative imaging, the majority of LGGs were preferentially resected. There were no differences in the tumor stratification, with WHO II oligodendrogliomas and oligoastrocytomas found in hospital B and WHO II astrocytomas found in hospital A. Other variations between centers included tumor characteristics (maximal diameter, eloquent location) and postoperative treatments, such as chemotherapy and radiotherapy. The authors further explored whether these factors or other variables might have contributed to the demonstrated survival differences. In a multivariate analysis, the relative hazard ratio (HR) was 1.8 (95% confidence interval (CI), 1.1 to 2.9, P value 0.03) for patients treated at a center favoring biopsy and watchful waiting, thereby confirming the survival benefit of early resection even without possible confounders (Jakola 2012).
In terms of QoL, Englot and colleagues performed a quantitative, comprehensive systematic literature review of seizure-control outcomes in 1181 patients with epilepsy across 41 studies after surgical resection of low-grade temporal lobe gliomas and glioneuronal tumors. Again, no RCTs were identified, however based on observational case series, gross total lesionectomy of temporal lobe LGGs resulted in improved seizure control over subtotal resection (Englot 2012). Nevertheless, these non-randomized studies (NRSs) are limited by selection bias. Patients who received total resection often have other desirable prognostic factors such as small non-eloquent tumors in otherwise healthy patients. Conversely, those undergoing subtotal resection or biopsy, or both, often have larger eloquent tumors and are less healthy patients. These confounding effects on survival comparisons can only be reliably obviated through randomization and proper trial methodology.
Historically, biopsy has been utilized for lesions located within or adjacent to eloquent parenchyma. Though surgical biopsy is a relatively safe procedure, some studies have suggested the risks of morbidity and inaccurate diagnosis may not outweigh the benefits. In a retrospective review of 81 patients with radiographic evidence of glioma who underwent stereotactic biopsy followed by surgical resection (within 60 days) between 1993 and 1998, diagnosis based on biopsy or resection differed in 40 cases (49%) (Jackson 2001). This was reduced to 30 cases (38%) when biopsy slides were reviewed preoperatively by three neuropathologists. The diagnostic accuracy in the Jackson study was lower compared to previously reports in the literature (63% to 95%). The authors attributed the high rates of diagnostic discrepancy to tumor heterogeneity, as there was a strong trend towards higher-grade malignancy on openly resected specimens. Following stereotactic biopsy, three patients (3.7%) had major complications and one (1.2%) had minor neurologic complications. Complications included cerebral hemorrhages, persistent hemiparesis, and temporary unilateral leg weakness. The authors concluded that stereotactic biopsy may actually be an unnecessary procedure in the management of suspected LGG, due to the high risk of inaccurate diagnosis subsequently delaying or negatively hindering appropriate surgical management.
In the absence of robust RCTs evaluating the initial steps of glioma diagnosis, physicians should consider prognostic factors when determining diagnostic and treatment methodologies for patients with LGGs. In one study by Pignatti, Cox analysis of 288 adults with LGGs demonstrated that patient age greater than 40 years, astrocytoma histology, large tumor diameter greater than 6 cm, midline extension, and preoperative neurologic deficit are unfavorable prognostic factors for survival (Pignatti 2002). Chang and colleagues has proposed a preoperative scoring system to prognosticate degree of lesion resectability, PFS, and OS in patients with LGG based on location, the Karnofsky Performance Scale (KPS), patient's age, and tumor diameter (Chang 2008). In clinical practice, patients can be divided into low- and high-risk subgroups based on their total number of unfavorable prognostic factors. One possibility yet to be studied is the hypothesis that immediate resection may decrease mortality in the high-risk subgroup, while conservative management with initial biopsy may be more appropriate in the low-risk subgroup. However, whether patients with low- or high-risk status would benefit from histological diagnosis via biopsy versus resection has yet to be definitively established.
Though not examined within the scope of this review, an alternative expectant management may improve QoL measures by eliminating the risks associated with biopsy, resection, or additional interventional therapy. In one small clinical trial, Recht and colleagues studied 26 patients who presented with a transient event (often seizures) and radiographic evidence suggestive of a low-grade primary supratentorial neoplasm who chose to withhold from all therapy until deemed necessary (Recht 1992). This group was compared to a similar group of 20 patients who received immediate intervention. With a median follow-up of 46 months, no identifiable difference between the groups in terms of survival or QoL was found. This evidence supports a personalized approach in determining a treatment plan for each patient presenting with LGG. While some patients may benefit from early knowledge of their histological diagnosis via biopsy or resection, others may prefer to postpone intervention unless it becomes necessary to maintain QoL. A larger clinical trial could be performed to provide more definitive evidence for observational treatment.
Importantly, the excluded studies did provide evidence in a non-randomized series that extensive resection yielded better outcomes (Ahmadi 2009; Claus 2005; Smith 2008). Region-of-interest analysis of 216 adults who underwent surgical resection of LGG showed that patients with at least a 90% extent of resection (EOR) had five- and eight-year OS rates of 97% and 91%, respectively (Smith 2008). In contrast, patients with less than 90% EOR had five- and eight-year OS rates of 76% and 60%, respectively. Analysis of 130 cases of adult supratentorial LGGs similarly demonstrated that both extended surgery and re-surgery were found to prolong OS and PFS (Claus 2005). Though there is evidence that increased EOR yields increased OS rates, it is still unknown whether biopsy without resection may similarly yield increased OS rates in certain cases.
Finally, there are recent data suggesting that an extended resection with a margin beyond MRI-defined abnormalities, a 'supratotal' resection, might improve outcomes in patients with LGG. Yordanova and colleagues enrolled 15 right-handed patients with a total of 17 WHO Grade II gliomas involving non-functional areas within the left dominant hemisphere all of whom underwent awake craniotomy with resection extended until cortical or subcortical "eloquent" areas as defined by intra-operative electrical mapping. Supratotal resection was achieved in 15/17 tumors based on postoperative MRI (with resection cavity > 10 cc larger than tumor volume). At mean 36 months follow-up, 4/15 patients experienced recurrence, 0/15 experienced anaplastic transformation, while a control group of 29 patients who underwent only complete resection had anaplastic transformation, seen in 7/29 cases (Yordanova 2011). These results resonated with the work by Duffau, Lang, and Bello, among others, and will continue to be an area of focus going forward.
Overall, this review demonstrated the continued need for RCTs to be performed in the area of biopsy versus resection for management of LGGs. Such a study will provide survival and QoL outcome data to guide management of patients with LGGs. Specifically, cases where an extensive resection can be completed safely may not necessarily need to be the focus of an RCT since the controversy in the field mainly relates to cases of subtotal resection.
The author team would like to thank Clare Jess and the editorial team for the Cochrane Gynaecological, Neuro-oncology and Orphan Cancer Group for their enthusiasm, support and guidance throughout the process. Peer reviewers Kathie Godfrey, Ruth Garside, Robin Grant, Andy Bryant, and Florence L-D are also commended for their insightful comments. We would like to thank Cassie Ludwig for her contribution to literature search, review, and appraisal in the original review.
This project was supported by the National Institute for Health Research, via Cochrane Infrastructure funding to the Cochrane Gynaecological, Neuro-oncology and Orphan Cancer Group. The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the Systematic Reviews Programme, NIHR, NHS or the Department of Health.
Appendix 1. Functionally independent survival scales
Karnofsky Performance Scale (KPS)
The KPS score runs from 100 to 0, where 100 is perfect health and 0 is death (Karnofsky 1949; Karnofsky 1951):
100% - normal, no complaints, no signs of disease
90% - capable of normal activity, few symptoms or signs of disease
80% - normal activity with some difficulty, some symptoms or signs
70% - caring for self, not capable of normal activity or work
60% - requiring some help, can take care of most personal requirements
50% - requires help often, requires frequent medical care
40% - disabled, requires special care and help
30% - severely disabled, hospital admission indicated but no risk of death
20% - very ill, urgently requiring admission, requires supportive measures or treatment
10% - moribund, rapidly progressive fatal disease processes
0% - death
Appendix 2. CENTRAL search strategy
#1 MeSH descriptor Glioma explode all trees
#6 (#1 OR #2 OR #3 OR #4 OR #5)
#7 MeSH descriptor Surgical Procedures, Operative explode all trees
#9 Any MeSH descriptor with qualifier: SU
#12 (#7 OR #8 OR #9 OR #10 OR #11)
#13 (#6 AND #12)
Appendix 3. MEDLINE Ovid search strategy
1 exp Glioma/
6 1 or 2 or 3 or 4 or 5
7 exp Surgical Procedures, Operative/
12 7 or 8 or 9 or 10 or 11
13 randomized controlled trial.pt.
14 controlled clinical trial.pt.
17 clinical trials as topic.sh.
20 13 or 14 or 15 or 16 or 17 or 18 or 19
21 6 and 12and 20
mp=title, original title, abstract, name of substance word, subject heading word, unique identifier
Appendix 4. Embase Ovid search strategy
1 exp glioma/
6 1 or 2 or 3 or 4 or 5
7 exp surgery/
12 7 or 8 or 9 or 10 or 11
13 exp controlled clinical trial/
14 crossover procedure/
15 randomized controlled trial/
16 single blind procedure/
19 (crossover* or cross over* or cross-over).mp.
21 (doubl* adj blind*).mp.
22 (singl* adj blind*).mp.
26 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25
27 6 and 12 and 26
mp=title, abstract, subject headings, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword
Differences between protocol and review
Data extraction and management
For future versions of this review, if identified data from new randomized controlled trials (RCTs) are available, they will be extracted onto a pre-designed data collection sheet. We will also record the following information, as recommended in Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Author, year of publication and journal citation (including language)
Inclusion and exclusion criteria
Study design, methodology
total number enrolled
primary cancer type
Risk of bias in study (assessment of risk of bias in included studies)
Duration of follow-up
Outcomes including overall survival (OS), progression-free survival (PFS), functionally independent survival (FIS), local tumor control, cause of death, steroid requirement and adverse events
Assessment of risk of bias in included studies
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
Other possible sources of bias
Data on outcomes will be extracted as below
For time to event (e.g. OS, PFS, and time to tumor progression) data, 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 attempt to estimate them from other reported statistics using the methods of Parmar 1998.
For dichotomous outcomes (e.g. adverse events or deaths if it is not possible to use a HR), 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, in order to estimate a risk ratio (RR).
For continuous outcomes (e.g. QoL measures), we will extract the final value and standard deviation (SD) of the outcome of interest and the number of patients assessed at the endpoint in each treatment arm at the end of follow-up, in order to estimate the mean difference (MD) (if trials measured outcomes on the same scale) or standardized mean difference (SMD) (if trials measured outcomes on different scales) between treatment arms and its standard error.
Where 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. New data will be abstracted independently by two review authors (AV, BJ) onto a data abstraction form specially designed for the review. Differences between review authors will be resolved by discussion or by appeal to a third review author, if necessary.
Risk of bias in future RCTs and CCTs will be assessed using the following questions and criteria (see Chapter 8 of Higgins 2011). Funnel plots corresponding to meta-analysis of the primary outcomes will assess the potential for small-study effects such as publication bias. If these plots suggest that treatment effects may not be sampled from a symmetric distribution, as assumed by the random-effects model, we plan to perform further meta-analyses using fixed-effect models.
Was the allocation sequence adequately generated?
Yes (low risk of bias), e.g. a computer-generated random sequence or a table of random numbers
No (high risk of bias), e.g. date of birth, clinic identity (ID)-number or surname
Unclear (uncertain risk of bias), e.g. not reported
Assessment of blinding will be restricted to blinding of outcome assessors, since it would not be possible to blind participants and treatment providers to the different interventions.
Was knowledge of the allocated interventions adequately prevented during the study?
Was similar care provided to patients in the treatment and control groups other than the intervention of interest?
Yes (low risk of bias), e.g. both groups were followed on similar schedules of neurologic exam and brain imaging
No (high risk of bias), e.g. each group was followed according to different schedules
Unclear (uncertain risk of bias), e.g. not reported
Incomplete reporting of outcome data
We will record the proportion of participants whose outcomes were not reported at the end of the study.
Were incomplete outcome data adequately addressed?
Yes (low risk of bias), if fewer than 20% of patients were lost to follow-up and reasons for loss to follow-up were similar in both treatment arms
No (high risk of bias), if more than 20% of patients were lost to follow-up or reasons for loss to follow-up differed between treatment arms
Unclear (uncertain risk of bias) if loss to follow-up was not reported
Selective reporting of outcomes
Are reports of the study free of suggestion of selective outcome reporting?
Yes (low risk of bias), e.g. if review reported all outcomes specified in the protocol
No (high risk of bias), otherwise
Unclear (uncertain risk of bias), if insufficient information available
Other potential threats to validity
Was the study apparently free of other problems that could put it at a high risk of bias?
Measurement of treatment effect will be done with HR and RR, and with QoL measures; we will use the MD between treatment arms.
If future data are sufficient and clinically similar studies are available, the results will be pooled in meta-analyses. For time-to-event data, HRs will be pooled using the generic inverse variance facility of RevMan 5. For dichotomous outcomes, the RR will be calculated for each study and then pooled. For continuous outcomes, the MD between the treatment arms at the end of follow-up will be pooled if all trials measured the outcome on the same scale, otherwise standardized MD will be pooled. Random-effects models with inverse variance weighting will be used for all meta-analyses (DerSimonian 1986).
Measures of treatment effect
We will use the following measures of the effect of treatment.
For time-to-event data, we will use the HR, where possible.
For dichotomous outcomes, we will use the RR.
For continuous outcomes (e.g. QoL measures), we will use the MD between treatment arms.
Unit of analysis issues
Unit of analysis issues will be reviewed by two authors (AV, BJ) according to Higgins 2011 and differences will be resolved by discussion. These include reports where:
groups of individuals were randomized together to the same intervention (i.e. cluster-randomized trials);
individuals undergo more than one intervention (e.g. in a cross-over trial, or simultaneous treatment of multiple sites on each individual); or
there are multiple observations for the same outcome (e.g. repeated measurements, recurring events, measurements on different body parts).
Dealing with missing data
We will not impute missing outcome data. For the primary outcome, if data were missing or only imputed data were reported, we will contact trial authors to request data on the outcomes among participants who were assessed.
Assessment of heterogeneity
Heterogeneity between studies will be assessed by visual inspection of forest plots, by estimation of the percentage heterogeneity between trials which cannot be ascribed to sampling variation (Higgins 2003), and by a formal statistical test of the significance of the heterogeneity (Deeks 2001). If there is evidence of substantial heterogeneity, the possible reasons for this will be investigated and reported.
Assessment of reporting biases
Reporting biases will be reviewed and recorded by two review authors (AV, BJ).
if sufficient, clinically similar studies are available, the results will be pooled in meta-analyses.
For time-to-event data, HRs will be pooled using the generic inverse variance facility of RevMan 5.
For dichotomous outcomes, the RR will be calculated for each study and then pooled.
For continuous outcomes, the MD between the treatment arms at the end of follow-up will be pooled if all trials measured the outcome on the same scale, otherwise SMD will be pooled.
Random-effects models with inverse variance weighting will be used for all meta-analyses (DerSimonian 1986).
Subgroup analysis and investigation of heterogeneity
Factors such as age, tumor size, tumor histology, extent of tumor resection, and length of follow-up will be considered in interpretation of any heterogeneity.
Determination of whether sensitivity analysis will be required will be determined by two review authors (AV, BJ) and differences resolved through discussion according to Higgins 2011. If sensitivity analyses were required, one will be performed that will exclude trials which do not report adequate concealment of allocation or blinding of outcome assessor.