Although 5 mm has been considered the historic standard within the oral cavity, this is not based on level 1 evidence. Herein we present the largest Canadian study analysing surgical margins of the oral cavity. Our findings are consistent with other recent studies [12,13,14,15,16,17], which suggest 5 mm margins may not be necessary.

In a recent survey of the American Head and Neck Society, 56% of respondents classified a clear margin as > 5 mm on microscopic evaluation [18]. Although this consensus is held true for most head and neck surgeons, there has been a strong push within the literature to redefine the definition for clear margin. In a study by Zanoni et al., [14] surgical margins of 381 patients with OCSCC were analyzed. Their findings indicate that patients with surgical margins ≤ 2.2 mm had significantly poorer LRFS, while those between 2.3 and 5.0 mm showed no significant difference. A similar investigation was conducted by Tasche et al. [15] In their retrospective review of 432 patients, they determined that there was a significant increased risk of recurrence with margins < 1 mm in close agreement with our study. Additional resection beyond 1 mm did not correspond to a significant difference in recurrence rates. Lastly, Bajwa et al [13] performed a retrospective, multicentre analysis (n = 669) assessing impact of surgical margins on LRFS, disease free survival (DFS), and OS. The results of their study revealed margins < 1 mm were associated with significantly poorer LRFS and DFS. These studies compliment the findings of the current paper well. Despite these findings, larger, multicentre analyses are still required for widespread adoption of new definition for clear margins. We are initiating such a multicentre study in collaboration with multiple other Canadian centres.

In the present study, multivariable analysis revealed age of diagnosis, tumour stage, pathologic nodal stage, and alcohol consumption per week were the strongest predictors of LRFS. Age of diagnosis, tumour staging, pathologic staging are all predictive factors that have been reported in previous studies [5, 6, 19]. Surprisingly, surgical margins were not retained on final analysis in our study. This finding contradicts the multivariable analysis of Bajwa et al. [13] and Zanoni et al. [14] We hypothesize that this may be due to lack of power. We will re-examine this as part of our future multicentre analysis.

Our study is not without limitations. Adjuvant therapy is a major confounding factor within our study, with 45% of patients undergoing radiation therapy post-operatively. Due to the retrospective design, it is not possible to determine the cause and effect relationship that this had on margin status, LRFS, RFS, and OS. Our study is also single centred and of modest sample size. Further multicentre analysis is required to validate our findings.