LAUSR

Editor We thank Bergamaschi et al. for their interest in our report on long-term outcomes for Dutch centres participating in the implementation phase of a structured training pathway for transanal total mesorectal excision (TaTME). The key message is that increased and multifocal local recurrences after TaTME for distal rectal cancer, as observed in Norway, have now been confirmed by centres implementing TaTME within a structured pathway1. A second key message is that in expert centres and centres that have progressed beyond their learning curve, this increased local recurrence risk seems to disappear2. This should act as a warning for colorectal surgeons thinking to start use of this complex technique without proper assessment and informed consent. We concur that the reported crude local recurrence rate of 10⋅0 per cent in the main cohort, comprising the first 10 patients from each of 12 centres, might be an underestimation. The complete cohort of 120 patients did not have extensive follow-up due to varying lengths of time between starting the pathway and completion of the tenth case. Kaplan–Meier analysis estimated a local recurrence rate of 3⋅4 per cent for 1 year and 10⋅4 per cent for 2 years. Indeed, the crude local recurrence rate of 15⋅0 per cent for the first 120 patients in a subgroup from four centres that underwent additional audit of their continued experience was higher than the 10⋅0 per cent for all hospitals, but this could be a coincidence. Univariate logistic regression identifies the risk of neglecting confounding factors, such as proctor presence during the first five cases. Owing to the limited sample size (denominator) and prevalence of the event, multivariate regression was not appropriate. The prevailing method for learning curve effects is a (risk-adjusted) cumulative sum assessment4. Cut-off values for the prolonged cohort were set at 10 parallel to the main study, and 40 based on cumulative sum control chart (CUSUM) analysis in other studies defining the learning curve for TaTME using perioperative outcomes such as duration of surgery, pathology, major morbidity and/or anastomotic leakage3–5. Quantification of the learning curve based on local recurrence by CUSUM requires an extensive consecutive series of patients, a reliable length of follow-up for oncological outcome and a large sample size if the incidence rate of the event is low. Waiting until sufficient follow-up data are available in the 12 centres would be unfeasible and in light of current uncertainty, unethical. Complex oncological surgery such as TaTME has been shown to have excellent results, but only in dedicated centres that have passed their learning curve2. New techniques for cancer surgery should be evaluated thoroughly in prospective trials. COLOR III is enrolling rapidly and quality assessment of the surgery together with reliable data on long-term outcomes will show the real value of TaTME. S. E. van Oostendorp1, C. Sietses3, R. Hompes2, M. Kusters1 and J. Tuynman1