Defining Non-inferiority Margins for Quality of Surgical Resection for Rectal Cancer: A Delphi Consensus Study.

INTRODUCTION: Quality of surgical resection metrics (QSRMs) have been used as surrogates for long-term oncologic outcomes in non-inferiority randomized clinical trials (RCTs) comparing laparoscopic and open surgery for rectal cancer. However, non-inferiority margins (ΔNI ) for QSRMs have not been previously defined.

METHODS: A two-round, web-based Delphi was used to define ΔNI for four QSRMs: positive circumferential resection margin (CRM), incomplete plane of mesorectal excision (PME), positive distal resection margin (DRM), and a composite of these outcomes. Overall, 130 international experts in rectal cancer (68 surgeons, 20 medical oncologists, 16 radiation oncologists, and 26 pathologists) were invited to participate. Experts were presented with evidence syntheses summarizing the association between QSRMs and long-term outcomes, and pooled quality of surgical resection outcomes for open surgery, and were asked to provide ΔNI for all outcomes balancing the risks and benefits of minimally invasive surgery.

RESULTS: Seventy-two experts participated: 57 completed the initial questionnaire and 58 completed the revised questionnaire, with 43 participating in both rounds. Consensus was reached for all individual QSRM ΔNI but not for the composite. The mean (standard deviation) ΔNI was an absolute difference of 2.33% (1.59%) for the proportion of positive CRMs when comparing surgical interventions for the treatment of rectal cancer: 2.85% (1.83%) for incomplete PME; 1.28% (1.13%) for positive DRMs; and 2.71% (2.28%) for the composite. However, opinions varied widely for the composite outcome.

CONCLUSIONS: Web-based Delphi processes are a feasible approach to generate ΔNI to evaluate novel surgical interventions. The generated ΔNI for QSRMs for rectal cancer can be used for future RCTs and non-inferiority meta-analyses.