The failure rates of treatment are high with locoregional recurrence rates between 40-80%.^[4] While tumour biology may dictate some of these, a surgical resection with attention to margins in three dimensions should reduce this risk. The deep resection margin in a carcinoma buccal mucosa often falls short of the clinically uninvolved cheek skin without consideration of the depth of invasion of the tumour. This may account for close margins in the depth.

The incidence of close and positive margins varies between 25.5% and 59%.^[5–8] It is not clear whether the margin status reported in literature includes deep margin along with mucosal or whether mucosal margins alone have been considered. This may have a role in the high local recurrence rates quoted for buccal carcinomas. Diaz et al quoted an incidence of invasion of buccinator of 34%. The locoregional recurrence rates were 46% and 41% for patients with and without muscle invasion, respectively. But there was no mention of the deep margin status.^[9]

Ota et al proposed an anatomical classification of buccal mucosal carcinoma based on depth of invasion on ultrasonography. They resected skin only when the tumour breached the buccinator muscle into the subcutaneous tissues and thus achieved a disease-specific survival rate of 73.7% and a local control rate of 89.5%.^[10] Liao et al reviewed their series of 331 patients with buccal carcinomas. All patients had a CT imaging to assess the distance of the skin from the tumour and if this was ≥ 13 mm, they would preserve the skin. This was considering the 10 mm surgical clearance and a minimum of 3 mm of skin to be retained to maintain its vascularity. They had a 5-year local control rate of 88%. Their deep margin was positive in <1% patients.^[11]

We have conducted a study in our institute combining ultrasound evaluation of depth and surgical management based on a surgical margin rather than an anatomical margin.

METHODS

The study was designed as a prospective study of all patients with carcinoma of the buccal mucosa. For purposes of comparison a retrospective cohort was analysed.

A prior Institutional Review Board (IRB) approval and institutional funding for the study was obtained (Ref:FG/8804/03/2014). All patients were counselled and explained the details of the study and informed written consent was obtained prior to inclusion in the study.

The prospective group (study group) comprised of all patients with carcinoma of the buccal mucosa attending the Out-Patient department (OPD) between March 2014 and June 2015. All patients with newly diagnosed, resectable squamous cell carcinoma limited to the buccal mucosa were included and those with obvious skin involvement (skin fixity, peau d’orange and ulceration) were excluded.

The preoperative investigations were directed towards staging and assessment of fitness for surgery. Clinical staging was done by direct examination of the lesion and cross-sectional imaging if the tumour was close to bone or retromolar trigone or if doubtful nodes were present. Chest X ray was done to rule out obvious lung secondaries.

All patients in the study group had a percutaneous ultrasound done to assess the maximum thickness of the lesion (A) and the minimum distance of the deepest aspect of the tumour from the skin (B). If the distance B was ≥ 13 mm then the overlying skin was preserved, and the line of excision passed through the subcutaneous tissue (Fig. 1) (At least 1 cm margin is needed for oncological clearance with an additional 3 mm of remaining skin to preserve subdermal vessels and ensure viability of overlying skin). If the distance ‘B’ was less than 13 mm, then the line of excision passed through the skin, thus removing it en bloc with the carcinoma (Fig. 2)

The specimen was then oriented with markers on the mucosal aspect and on the deep aspect of the lesion and sent for routine histopathological examination.

We had 45 patients of buccal carcinoma in the period from Jan. 2008 to Dec. 2010 as part of a retrospective data that we had collected. Out of these 15 patients (Retrospective group) fulfilled the inclusion and exclusion criteria (i.e. there was no skin involvement on physical examination) that was formulated for the present study and these alone were, also, included as a comparable group. All patients were followed up for a period of 24 months.

Statistical Analysis

The distribution of tumour stage, surgical procedures, histopathological factors, adjuvant therapy in the prospective and retrospective groups were compared using Chi-square and Fisher’s exact test. The Kaplan-Meier estimator was used to compare the disease-free survival in the two groups.

RESULTS

The male to female ratio in the study group was 2:1 and the male preponderance was seen in our retrospective group also (93.3%). The mean age of the study group was 60.6 yrs and 52.3 yrs in the retrospective series.

Surgery

All study group patients required skin excision for clearance of tumour as the distance of tumour to skin was < 13 mm. There was no case of skin excision in retrospective group as it was the practice to preserve skin with no clinical signs of involvement. There were significantly more patients who required flap reconstructions in study group than the retrospective group (Table 1).

Histopathology

In the study group, one patient had no tumour in the final post-surgical biopsy. He was staged as pT1 as the initial biopsy had shown early invasive squamous cell carcinoma. The pathological tumour and nodal staging were comparable between both groups (Table 2).

Histopathological Margin Status

In the study group, the mean distance of tumour from the skin (B, Fig. 1) as measured with ultrasound was 5.3 ± 2.5 mm. All patients had less than 13 mm distance of tumour from the skin and thus required excision of skin. However, in the study group, one patient did not have any tumour in the postsurgical biopsy specimen and hence deep margin status was assessed for 14/15 (93.3%) patients.

The deep margins were adequate (>5 mm) in 12/14 (85.7%) patients and this was a significant improvement over the retrospective group where negative margins were achieved in 13.3% (p<0.001)

In the study group, the mucosal margin was negative in 10/14 (71.4%) and close in 4/14 (28.5%) patients; and negative in 3/15 (20%), close in 8/15 (53.3%) and positive in 4/15 (26.6%) patients in the retrospective group. For the statistical analysis, the close and positive margins were combined as inadequate margins and the difference between the two groups was statistically significant (Fisher’s exact test, p=0.004). (Table 2)

Adjuvant Therapy

Six patients in the study required adjuvant radiotherapy for adverse histopathological factors; one required adjuvant chemotherapy in addition to radiotherapy. Four patients completed radiotherapy while two defaulted without taking a single dose of radiotherapy and one defaulted on the second dose of chemotherapy in the study group (Table 3). Significantly more patients required adjuvant radiotherapy and chemotherapy in the retrospective group.

Locoregional Recurrences

There were three patients with recurrences in the study group, two local buccal recurrence and one nodal recurrence. For purposes of analysis, we looked at locoregional recurrence as a composite outcome variable. Univariate analysis in the study group showed that factors like tumour grade, pathological tumour stage, pathological nodal stage, perineural invasion (PNI), extranodal spread (ENS), mucosal margin status and deep margin status were not statistically significant predictors of locoregional recurrence. (Table 4)

Univariate analysis in the retrospective group showed that prognostic factors like tumour grade, pathological tumour stage, pathological nodal stage, mucosal margin status and deep margin status were not statistically significant predictors of locoregional recurrence. (Table 4)

Follow Up

All the patients were followed up for at least 2 years or till an event occurred. The mean follow-up period was 20 months in the study group and 28.3 months in the retrospective group. Kaplan Meier survival curves comparing both groups for loco-regional recurrence were plotted and the groups did not differ significantly (Fig. 3; log rank test; p=0.838).