We will consider both ways, randomized and non-randomized studies which compare high tie (ligation at the aortic origin) versus low tie (ligation below the origin of the left colic artery) of the IMA for sigmoid or rectal resection for cancer. Furthermore, in order to be considered for inclusion, studies have to report outcomes for sigmoid (left colectomy) or rectal cancer surgery (anterior resection/sphincter-sparing surgery or abdomino - perineal resection) and to compare high tie versus low tie. We will not impose any language or publication status restrictions.

The studies will be excluded from analysis if the outcomes of interest will not report the two techniques or whether it will be not possible to extrapolate them from the published results, also studies will be about benign lesions or inflammatory bowel disease without a distinct group of patients with cancer.

Patients of any age and sex with sigmoid or rectal cancer will be considered.

Both sigmoid or rectal resection with high and low IMA tie.

The following outcomes will be observed:

A systematic search will be conducted in: Medline, Embase, Cochrane Central Register of Controlled Trials, CINAHL, BioMed Central, Science Citation Index and performed on all studies for potentially relevant trials comparing high with low IMA tie. A secondary search will be conducted reviewing unpublished literature databases including: Greynet, SIGLE, National Technological Information Service, British Library Integrated catalogue, Current Controlled Trials and the Cochrane Central Register of Controlled Trials

Combinations of the following search terms will be used: inferior mesenteric artery'; 'lymph node' or 'lymph nodes'; 'colon' or 'rectum'; 'cancer', 'neoplasia', 'tumour', or 'tumor'.

We will search the related article of Pub Med and all references.

To minimize retrieval bias we will perform a new manual search method that utilize the Google Scholar database and manually searched seven high-impact journals, chosen on the basis of the frequency of articles and on expert opinion.

The reference lists of all potentially eligible studies will be reviewed. Researchers who may have carried out relevant studies will be contacted. Animal trials will be excluded.

Two authors (RC and CB) will assess titles or abstracts of all identified studies independently and exclude all the irrelevant ones. Full text articles of potentially relevant studies will be obtained. These studies will be assess independently in an unblended standardized manner by the two authors (GDR and AS) as to whether they met the inclusion criteria for this review.

We will develop a data extraction sheet (based on the Cochrane Consumers and Communication Review Group's data extraction template), pilot tested it on ten randomly-selected included studies and refined it accordingly. One author (ST) will extract the data from the included study and the second author (GN) will check the extracted data. Disagreements will be solved through discussion, if necessary, by involving an independent third author (AR).

The following information will be extracted by one author (ST) for each included trial:

- Year and language of publication.

- Country in which the trial was conducted.

- Year of conduct of trial.

- Single-center or multicenter trial.

- Characteristics of trial participants

- Inclusion and exclusion criteria.

- All outcomes

Two authors (ST and EF) will perform the statistical analysis in line with recommendations from the PRISMA statement [²²] and the Cochrane Handbook for systematic reviews [²³]. Statistical analysis for categorical variables will be performed by using the odds ratio (OR) as summary statistic. This ratio represents the odds of an adverse event occurring in the high tie group compared with the low tie. The Mantel-Haenszel method will be used to combine the ORS for the outcomes of interest) [²⁴,²⁵]. For continuous variables statistical analysis, we will use the weighted mean difference (WMD). A negative WMD favored the high tie group and subgroups, and the estimated point of the WMD will be considered statistically significant with P < 0.05, if 95% confidence interval (CI) did not include the value zero.

Fixed effect models and Random effect model will be initially calculated for all outcomes. Then we will test the homogeneity among the studies by calculating the chi² and I². I² or "inconsistency" describes the proportion of total variation in studies and it is independent from the number of combined studies. If the test rejects the assumption of homogeneity of the studies, then it is not appropriate to use a fixed effect model, and random effect analysis will be reported. Sensitivity analyses will also lead to explore the statistical heterogeneity [²⁶].

If the studies report their continuous variables as medians with ranges that a meta-analysis cannot use, we will assume the mean to be equal to the median value itself and estimated the standard deviation (SD) as a quarter of the range (samples = 70) or one sixth of the range (samples > 70) [²⁷]. If neither range nor other measure of dispersion will be reported, it will be impossible to estimate the mean and the SD based on the published data and the corresponding continuous variables will be excluded from the statistical pool.

Statistical analysis will be conducted by using the statistical software Review Manager Version 5.0.

Two authors will assess the risk of bias of the trials independently (ST, JD) using for the RCTs the instructions given in the Cochrane Handbook for Systematic Reviews of Interventions and for CCTs the modified Newcastle-Ottawa scale [²⁸,²⁹]. Graphical exploration with funnel plots will be used to evaluate publication bias [³⁰].

A narrative synthesis of the included studies, risk of bias and results will be performed. If heterogeneity will present a I² < 50% will reported the outcome results using a random effects meta-analysis. We will conduct sensitivity analyses based on study quality.

The article will be submitted to a peer-reviewed journal.

University of Perugia

1 September 2011

1 December 2011