Study Eligibility Criteria
Studies selection will be performed according to the PICO criteria of PRISMA described below(39).
Types Of Studies
Randomized controlled trials (RCTs), including cluster RCTs, controlled (non-randomized) clinical trials (CCTs) or cluster trials, controlled before-after (CBA) studies, prospective and retrospective comparative cohort studies, and case-control or nested case-control studies will be included. Cluster randomized, cluster non-randomized, or CBA studies will be included only if there are at least two intervention sites and two control sites. Cross-sectional studies, case series, and case reports will be excluded.
Type Of Participants
We will include adult patients with CRCLM and with indication to perform LR or PVE, irrespective of the number of lesions or their localization. At least two volumetric estimations of FLR, one before and one after the procedure (LR or PVE) will be required for the inclusion of the study in the current review. The usual indication to perform PVE is an insufficient FLR. Accordingly, a separate analysis for LR and PVE indications will be performed.
Type Of Interventions
In order to achieve the primary objective, the type of intervention to be taken into account will be neoadjuvant chemotherapy, irrespective of type, number of cycles or other characteristics. According to the National Cancer Institute, neoadjuvant chemotherapy is defined as “treatment given as a first step to shrink a tumor before the main treatment, which is usually surgery”(40).
In order to achieve the secondary objectives, we will perform sub-analysis for the following type of interventions:
- Type of chemotherapy –addition of bevacizumab to the chemotherapy regimen.
- Number of cycles: the intervention will be administration of more than 6 cycles of neoadjuvant chemotherapy. Depending on the data found in the selected studies, this number may be changed.
- Time between the end of chemotherapy and the procedure (LR or PVE): the intervention will be considered when this wait is less than 8 weeks. Depending on the analysis performed in the selected studies, this number may be changed.
For the primary objective, the control group will include patients without chemotherapy submitted to the procedure (defined as LR or PVE). We will accept studies in which the control group contains patients with benign disease or other types of neoplasia. However, these studies will be carefully analyzed to detect possible biases of selection.
For the secondary objectives, the control group can include patients with chemotherapy that do not meet the type of intervention mentioned above, patients without chemotherapy (similar to primary objective) or both.
- Liver regeneration/hypertrophy rate: future liver remnant regeneration rate (FLR3), calculated as follows:
Where FLRf is the final future liver remnant (after the procedure – LR or PVE) and FLRi is the initial future liver remnant (before the procedure). The timing of FLRf will depend on the data reported in the selected studies, but we expect mainly volumetric data in the first month after the procedure, giving that the most part of the hypertrophy occurs in this interval.
- Surrogate volumetric data:
- Total liver volume changes.
- Changes in the volume of the liver to be resected (derived from pre- and post-procedure volumes in the case of PVE and pre-procedure volumes and weight of the resected liver in the case of LR).
- Postoperative liver failure or liver dysfunction. Accepted definition of liver failure will be:
- 50-50 criteria as proposed by Balzan – association of a prothrombin time of less than 50% and serum bilirubin of more than 50 μmol/L on 5th postoperative day(16).
- Bilirubin peak of more than 120 μmol/L during the postoperative period of a major hepatectomy(41).
- Grade B and C of post-hepatectomy liver failure according to International Study Group of Liver Surgery(42).
Review eligibility criteria
We will include studies independently of geographic location or year of publication. We will accept unpublished material and abstracts from congresses. There will be no language restrictions. Editorials, letters or commentaries will be excluded during the screening of titles and abstracts. Experimental studies on animals will also be excluded.
Literature search strategies will be conducted using medical subject headings (MeSH) and text words related to the objectives of this systematic review. We will search the following electronic databases:
- PubMed (1990 to present)
- Scopus (1990 to present)
- Web of Science (1990 to present)
- Embase (1990 to present)
- Cochrane Central Register of Controlled Trials (1996 to present)
The key words used to perform the search in each electronic database are shown in Appendix 1.
Before proceeding to write the manuscript draft, the search of the literature will be updated, in order to identify any new publication which could be relevant to the objectives of this systematic review.
Searching Other Resources
Reference lists of all primary studies and review articles will be manually searched for additional references. Authors of identified studies will be contacted if needed in order to ask them to identify other published and unpublished studies. Errata or retractions from eligible studies will be searched on PubMed and the date this was done will be reported in the review.
Data Collection and Analysis
Literature search will be loaded in a specially created Mendeley folder in order to access to titles and abstracts and will be listed in a specially created Excel file with the following coding: included, not included and 2nd look. Several methods will be used to identify duplicate publications: juxtaposing author names and comparing sample sizes, outcomes and text of the abstract.
Two review authors (MP and RC) will independently screen titles and abstracts of all the potential studies identified as a result of the search, and code them as “included”, “not included” or “2nd look”. Full text of study reports coded as “included” or “2nd look” will be retrieved and two review authors (MP and RC) will independently analyze the full text, identifying studies for inclusion and recording reasons for exclusion of the ineligible studies. Study authors will be contacted when additional information will be needed to resolve questions about eligibility. Any disagreement between the two authors will be solved by a 3rd reviewer (RJ). Duplicates and collate multiple reports of the same study will be identified and excluded, thus each study rather than each report will be the unit of interest in the review. The selection process will be recorded in sufficient detail to complete a PRISMA flow diagram and a table of excluded studies features.
Data collection process
An Excel-based data collection form will be used to register study characteristics and outcome data. The data collection will be piloted on five studies previously identified as significant for this review. Two review authors (MP and RC) will independently extract study characteristics and outcome data from included studies. Any disagreement between the two authors will be solved by a third reviewer (RJ). To ensure consistency across reviewers, calibration exercises will be conducted before starting the review. Study corresponding authors will be contacted to resolve any uncertainties (three e-mail attempts at maximum). In order to extract data not reported in a numeric format, graphically presented data will be translated into usable format using OriginPro v 7.5 from OriginLab. Before final revision of the draft, another search will be performed in order to identify duplicate publications among the selected articles, following the same previously described methodology.
The following study characteristics and outcomes will be extracted:
- Methods: study design, total duration of study and run-in period, number of study centers and location, study setting, withdrawals, date of study.
- Participants: number, mean age, age range, gender, inclusion criteria, exclusion criteria.
- Interventions: intervention, comparison, and any cointerventions.
- Outcomes: primary and secondary outcomes specified and collected, time points reported.
- Notes: funding for trial, notable conflicts of interest of trial authors.
Outcomes And Prioritization
The main outcome of this systematic review will be FLR3, which represents the hypertrophy of the remnant liver after surgery. This outcome is derived from the estimated volume of the remnant before and after the intervention (LR or PVE). Even though both interventions induce liver hypertrophy, presumably FLR3 after LR or PVE will not be comparable. Therefore, FLR3 will be analyzed separately after each one of the interventions. FLR3 data will be expressed as mean ± SD. If data is offered in other forms (median – range or median – IQR), mean ± SD will be calculated following the recommendations of Cochrane Handbook for Systematic Reviews of Interventions, whenever possible.
Regarding the timing of post-procedure volume calculation, homogenous data are expected in PVE studies. The majority of preoperative PVE protocols plan surgery four weeks after the embolization procedure. However, the timing of remnant volumetry after LR could vary between studies. The majority of included studies most likely perform volume calculation one month after the procedure, when 80% of liver hypertrophy has occurred. Whenever remnant volume has been calculated at several time points, the one obtained at one month after LR will be chosen. The time of post-procedure volumetry will be registered for each study.
The secondary outcome will be liver failure/dysfunction as defined above. This outcome will be calculated only for patients submitted to LR. This outcome will try to determine whether there is a correlation between a possible lower hypertrophy rate in patients with neoadjuvant chemotherapy and subsequent liver dysfunction rate.
Assessment of bias
Two investigators (MP and RC) will independently assess risk of bias for the included studies. Risk of bias will be assessed by the ROBINS-I tool for non-RCT studies(43). In this tool, risk of bias is assessed within specified domains, including (1) bias due to confounding, (2) bias in selection of participants into the study, (3) in classification of interventions, (4) bias due to deviations from intended interventions (5) bias due to missing data, (6) bias in measurement of outcomes, (7) bias in selection of the reported result, and (8) overall bias. Since assessments are inherently subjective and there are no strict and objective criteria to judge bias within the ROBINS-I tool, disagreements will be resolved via discussion between the two investigators or by the intervention of a third (RJ).
The quality of each selected article will be assessed by one reviewer and verified by another. The quality of evidence will be determined with the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system(44).
If selected studies are sufficiently homogeneous in design and comparators, we will perform a meta-analysis of reported results.
Measures Of Treatment Effect
Dichotomous data (liver failure or dysfunction as previously defined or presence of ascites, encephalopathy) will be analyzed by using risk ratio (RR) with 95% confidence interval (CI) and continuous outcomes (FLR3 or surrogate volumetric data - total liver volume changes and changes in the liver volume to be resected) as mean difference or standardized mean difference when different scales are used (e.g. FLR3 vs surrogate volumetric data).
If a study is suspected of comprising skewed data, this is commonly indicatedby reporting medians and interquartile ranges. When this is found, transformations to mean differences will be carried out. If this is not possible due to lack of data, the data will be considered as skewed. If the data are skewed, a meta-analysis will be not performed, though a narrative summary will be provided instead.
Unit Of Analysis Issues
The unit of analysis will be individual participants affected by liver metastasis and candidates for LR or PVE. If any cluster-randomized studies are unexpectedly found, the data will be included in the analysis if the results are adjusted for intra-cluster correlation. If any cross-over randomized studies are found, the data prior to the cross-over will be included. When a study has more than two treatment groups, the additional treatment arms will be presented. Where the additional treatment arms are not relevant, they will not be taken into account.
Dealing With Missing Data
Investigators or study sponsors will be contacted to verify key study characteristics and obtain missing numerical outcome data (e.g. when a study is presented as abstract only). If this information is not available from the study authors, it will be obtained, where feasible, by using calculations provided in the Cochrane Handbook for Systematic Review Interventions. The impact of including such studies will be assessed in a sensitivity analysis. If we are unable to calculate the standard deviation from standard error, interquartile range, or P values, we will impute standard deviation as the highest standard deviation in the remaining studies included in the outcome.
Assessment Of Heterogeneity
Clinical heterogeneity will be tested by considering the variability in participant factors among trials (for example age) and trial factors (randomization concealment, blinding of outcome assessment, losses to follow-up, treatment type, co-interventions). Statistical heterogeneity will be tested using the Chi2 test (significance level: 0.1) and I2 statistic (0% to 40%: might not be important; 30% to 60%: may represent moderate heterogeneity; 50% to 90%: may represent substantial heterogeneity; 75% to 100%: considerable heterogeneity). If high levels of heterogeneity among the trials exist (I2 >=50% or P <0.1) the study design and characteristics in the included studies will be analyzed. The source of heterogeneity by subgroup analysis or sensitivity analysis will be explained.
Each outcome will be calculated using the statistical software RevMan 5.1, according to the current version of the Cochrane Handbook for Systematic Reviews of Interventions. The Mantel-Haenszel method will be used for the fixed effect model if tests of heterogeneity are not significant. If statistical heterogeneity is observed (I2 >=50% or P <0.1), the random effects model will be chosen. Data will be presented in text and tables, in order to summarize the characteristics and findings of included studies. The analysis will describe the findings and associations within individual studies as well as among all the studies included in this review.
Subgroup Analysis And Investigation Of Heterogeneity
Subgroup analysis will be used to investigate possible sources of heterogeneity, based on the following parameters:
- General characteristics of included patients (age, sex)
- Timing of post-procedure volumetry
- Type of procedure (hepatectomy vs PVE)
Sensitivity analysis will be performed to explain the source of heterogeneity:
- Analysis of the material retrieved (full text vs abstract only, preliminary data vs final results, published vs unpublished material)
- Risk of bias (performing analysis by omitting studies evaluated as of high risk of bias)
If there is a need to amend this protocol, the date of each amendment will be registered, describing the change and giving the rationale in this section. Changes will not be incorporated into the protocol.
Conclusions will be based on findings from the quantitative or narrative analysis of the studies included it this review. We will avoid making recommendations for clinical practice but we will focus on the remaining uncertainties in the field and the need for future clinical investigation.