Oral Cavity Cancer: Margin Management

Hi Team,

Hope everyone had a good weekend. Week two of oral cavity, which means this week is margins, and we close out the topic next week with management of the mandible and adjuvant therapy. Same as always, Wednesday mornings @ 7AM sharp(ish), with donuts for those who can join in person. If Wednesday does not work for you, email me or grab any of our head and neck surgeons while you are in the OR.

Margins are the part of an oral cavity resection you have the most control over and the least feedback on. The AHNS curriculum gives you exactly two papers here, Maxwell 2015 and Zanoni 2017, and they answer two different questions: where the margin comes from, and how wide it has to be. Both have picked up meaningful updates in the last two years, so we will read the classics and then look at what has changed.

Where the Margin Comes From - Specimen Versus Tumor Bed

Maxwell JAMA Otolaryngol 2015 is a five-institution retrospective series of 280 patients with pT1-T2 pN0 oral tongue squamous cell carcinoma, grouped by how the surgeon sampled the intraoperative margin. Three-year local recurrence-free survival was roughly 90% when no tumor bed sampling was done, versus roughly 80% when margins came primarily from the tumor bed (p=0.03). Positive glossectomy margins were reported in 7.7% of the specimen-only group and 24.2% of the tumor bed group. Read it knowing that surgeons who reach into the wound bed may be the surgeons operating on the harder cases, so selection bias is likely driving a lot of these differences. It is also an interesting example of pooling five centers to gain "generalizability" while creating collinearity between a critical variable, where the margin is taken from, and the institution and surgeon taking it.

Caraway Otolaryngol Head Neck Surg 2026 is the meta-analysis that puts numbers on it: nine studies, 1,240 patients. The specimen-based approach produced a higher intraoperative re-resection rate, a significantly lower final positive margin rate, and less local recurrence. It did not change overall recurrence or survival. Our practice at Penn is specimen-driven, with the surgeon orienting the specimen and sampling it with pathology rather than sending separately labeled wound bed biopsies.

How Wide is Wide Enough - 5mm, 2.2mm, and Specimen Shrinkage

Zanoni JAMA Otolaryngol 2017 is 381 oral tongue patients from Memorial Sloan Kettering, and it is the paper that challenged the 5mm rule. The threshold that actually separated outcomes was 2.2mm. Two-year local recurrence-free survival was 91.8% for margins over 5.0mm and 93.5% for margins of 2.3 to 5.0mm, statistically indistinguishable. Below 2.2mm the hazard ratio for local recurrence was 2.00, and a frankly positive margin was 5.73. TL;DR: the gap between 3mm and 6mm on the slide may not be the gap you think it is, but the gap between 1mm and 3mm is.

Here is the part that changes what you do in the room. Johnson Head Neck 1997 measured the same margins at multiple points through processing and found the lingual mucosal margin shrinks 30.7% from resection to final microscopic assessment, the deep tongue margin 34.5%, and the labiobuccal mucosal margin 47.3%. Most of that happens the instant the specimen comes off. Their recommendation was to resect 8 to 10mm in vivo to end up with 5mm on the slide. When you are standing at the tongue with the marking pen, that is the number that matters, not the one on the pathology report. To be critical, this is from 30 years ago using data from 45-50 years ago. An additional centimeter margin in a glossectomy is the difference between a partial gloss and hemigloss or hemigloss/subtotal or subtotal/total. Surgical technique, anatomy, and tissue handling is critical in these settings. This is where it's helpful to understand the intrinsic and extrinsic tongue muscles, fascial planes in the tongue, and location of the lingual artery bilaterally.

Hung Ann Surg Oncol 2025 adds a variable I think we underuse. Working from 7,420 Taiwan Cancer Registry patients who all had negative margins, they built a margin-to-depth-of-invasion ratio and split at 0.5. Patients below 0.5 had 3-year disease-free survival of 74% versus 86%, and overall survival of 79% versus 89%, despite receiving more postoperative radiotherapy. A 4mm margin on a tumor with 3mm depth of invasion is not the same margin as 4mm on a tumor with 12mm depth of invasion, and our reporting treats them identically. Worth asking yourself: how many other variables are we flattening into a single number because the single number is easier to say at tumor board?

Clearing a Positive Margin - Does the Re-Resection Rescue You?

This is the one I most want you to have an opinion about before Wednesday. Bulbul Otolaryngol Head Neck Surg 2019 pooled eight studies and compared three groups: initially negative margins, initially positive margins revised to negative, and persistently positive margins. Against initially negative margins, the revised group had a hazard ratio for local recurrence-free survival of 2.897 and the persistently positive group 3.795, both highly significant. Revising a positive margin to clear does not buy back the outcome of a patient who was negative on the first pass, and it only trended better than leaving it positive.

Several explanations are on the table and they are not mutually exclusive: the tissue you go back for is not reliably the tissue that was positive, the biology that produces a positive margin is the same biology that produces recurrence, and a re-resection often reads as negative because you missed the right spot a second time. None of this is an argument against re-resecting. It is an argument for getting it right the first time, which is the entire case for specimen-driven assessment above (and for choosing the right operation out of the gate).

Bone and Deep Margins

Shaw Head Neck 2004 prospectively studied 100 consecutive mandibular resections for untreated oral cavity squamous cell carcinoma in Liverpool, UK, 65 segmental and 35 marginal, with 62% of mandibles showing tumor invasion. Local recurrence was 21% and five-year disease-specific survival was 68%. The finding to carry with you is that infiltrative bone invasion behaved worse than erosive invasion, and that soft tissue factors still dominated prognosis. Bone invasion earns the T4a designation, but the mandible is frequently not the thing driving the outcome.

Nair Indian J Surg Oncol 2025 is a recent review out of Tata Memorial Hospital in India proposing a classification of marginal mandibulectomy by resection plane and mandibular location, and arguing that for tumors abutting the mandible without bone invasion, marginal mandibulectomy gives comparable oncologic outcomes with substantially better function. Good setup for next week when we take the mandible on properly.

Adjuvant Therapy for the Close Margin, and Where This Goes Next

Kohler Eur Arch Otorhinolaryngol 2026 took 375 oral cavity squamous cell carcinoma patients with margins between 1 and 5mm and asked who actually benefits from adjuvant treatment. Stratified by pathologic risk, low-risk patients saw no significant benefit, medium-risk patients benefited from either radiotherapy or chemoradiotherapy, and high-risk patients benefited most from chemoradiotherapy. This is retrospective, so do not treat it as guideline, but it frames the right question: a close margin is a risk factor, not a treatment plan by itself.

For the informatics-minded among you, Michcik Front Med 2025 is a ten-patient pilot from Gdansk, Poland, in which the resected specimen is 3D scanned and the intraoperative margin sample sites are mapped virtually, so the final report can state the width the surgeon actually took. In one case a reported R1 margin of 2.5mm turned out to be 5.5 to 7.0mm once the intraoperative sample was accounted for, and the tumor board took that patient off adjuvant radiotherapy. Ten patients is ten patients, and this needs far more validation. It is a direct descendant of the 3D scanning and printing work that Mark Urken's group and now Mike Topf out of Vanderbilt have been doing for years, and it points at something real: a meaningful share of our margin disagreements are documentation problems rather than surgical ones.

Thanks for taking the time to read this, and looking forward to arguing about the 2.2mm number on Wednesday!

Have a great week!

Rob