Transanal Versus Laparoscopic Total Mesorectal Excision For Mid And Low Rectal Cancer. A Multicentre Randomised Clinical Trial on Anastomotic Leak

This study is designed as a prospective, multi-center, randomized, open-labelled, parallel group, non-inferiority trial. The patients will be randomized and assigned to transanal mesorectal excision (TaTME) or laparoscopic mesorectal excision (LpTME) at a ratio 1:1, based on a centralized computer generated, blocked randomization list stratified by center. Six centers will be involved. Within each center a block size of four will be used, stratified by gender and age (±3 years). A central electronic data capture system will be generated by the promoting center for data collection. All the participating surgeons and centers will be required to document their previous experience with TaTME (a minimum of 10 documented procedures is required). Moreover, they will be invited to submit a video showing their surgical technique in performing a TaTME operation (said video must contain no reference to patient ID, and is used exclusively to demonstrate the participating surgeon's technique). The video will be reviewed by the promoting center in order to standardize the procedure, prior to the beginning of the recruitment process. In each center, before starting effective recruitment of patients to the trial, the surgical team will be required to perform a minimum of 4 TaTME operations based on the adopted standardized procedure to calibrate the participant surgeons.

All statistical analyses will be conducted on an "intention to treat" basis.

Statistical methodology The trial is designed as a non-inferiority trial. In statistical terms, the study will establish whether the anastomotic rate (%) for the experimental treatment is not worse than the anastomotic rate (%) for the standard treatment by more than a specific non-inferiority margin, δ, set to 0.05, with δ denoting the difference anastomotic rate rates (experimental vs. control). Formally, the study will test the one-sided null hypothesis H0: δ > = 5% vs. the alternative H1: δ < 5%, and will reject the null hypothesis with a 5% level of significance if the upper bound of a standard asymptotic 90% confidence interval for difference in proportions is below the 5% non-inferiority margin.

Sample size In recent studies (1-5) estimating the anastomotic leak ("failure") rate (%) in the standard LpTME (control technique) ranged from 6.4 to 18.8 % (median of studies, 13%), while in the two recent studies evaluating TaTME (experimental technique) it was 6.7% and 8.6 %, respectively, (resulting in a weighted rate of 7%). The investigators therefore hypothesised a failure rate of 13% (that equivalently means a "success" rate of 87%) in the control arm and a crude (relative) 6% (4.6%) reduction in the experimental arm (that equivalently means a "success" rate of 93%) over the standard arm. Based on the non-inferiority hypothesis and the above findings, and assuming a non-inferiority limit of 0.05, a total of 184 patients (92 for each arm) is required i.e., if there is a true difference in favour of the experimental treatment of 6%, then 184 patients are required to be 80% sure that the upper limit of a one-sided 95% confidence interval (or equivalently a 90% two-sided confidence interval) will exclude a difference in favour of the standard group of more than 5%. The inclusion period will be of 2 years. The results will be available within 3 years of starting recruitment. The investigators will consider still acceptable and not different clinically a failure rate of 18% (i.e., a non-inferiority limit of 5%) as this value is below the 18.8% maximum failure rate estimated in the standard LpTME. Additionally, the non-inferiority limit of 5% has been also recently indicated in a similar trial (ETAP-GRECCAR 11 TRIAL) (assuming conversion to open as "failure" event) registered at clinicalTrials.gov, NCT02584985.

Statistical analysis Both per-protocol and intention-to-treat analysis will be performed, as currently recommended in literature. Descriptive data for continuous variables will be reported as mean (standard deviation [SD]) or median (25th and 75th percentiles), as appropriate. Descriptive data for categorical variables will be reported as number of observations (percentage). For continuous variables, univariate comparison between LpTME and TaTME will be performed by the Student's t test for normally distributed data (based on the Kolmogorov-Smirnov test) or the Mann-Whitney U test for non-normally distributed data. The chi-square test, or the Fisher exact test, as well the Kruskal-Wallis test, as appropriate, will be used for comparing discrete variables. Logistic regression models will be additionally carried out and survival analysis of postoperative binary outcomes, as appropriate. All the statistical tests will be performed by using the SPSS statistical package, version 20.0 (SPSS Inc., Chicago, IL) for Windows (Microsoft, Redmond, WA).