Accuracy of frozen sections in assessing margins in oral cancer resection

J Oral Maxillofac Surg 55:663-669, 1997

Accuracy of Frozen Sections in Assessing Margins in Oral Cancer Resection ROBERT

A. ORD” AND SEENA AlSNERt

Purpose: This study examined the accuracy of frozen section diagnosis of tissue samples from surgical margins compared with the final histologic diagnosis of the same tissue. The total resection specimen was also examined to see whether frozen sections were helpful in predicting negative margins for the entire cancer. The nature of positive and negative margins and their implications for the surgeon are discussed. Patients and Methods: The records of 49 consecutive patients with previously untreated squamous carcinoma of the mouth were reviewed. All frozen and permanent sections were evaluated by one pathologist. Margins involved by carcinoma, carcinoma in situ, dysplasia, or with carcinoma within 5 mm were defined as positive. Histologic findings were compared with the patient’s clinical course to define the relationship between positive margins and local recurrence. Patients were followed for 17 to 45 months or until death. Results: Three hundred four of 307 frozen sections showed concordance with the permanent section of the same tissue sample (two false negative and one false positive), an accuracy rate of 99%. When the final margins of the resected surgical specimen were compared with the frozen section diagnoses, ten patients had positive final margins. In three patients, these were diagnosed by frozen section. Seven patients had final margins that were positive when the surgical resection specimen was examined but were not diagnosed by frozen section. A greater local recurrence note was found in patients with invasive carcinoma at the margin, dysplastic margins, and margins within 5 mm of the cancer. Conc/usions: Although frozen sections are extremely accurate, they are not as reliable in eliminating positive margins in the final specimen as the surgeon might hope.

The best chance of surgical cure in oral cancer is complete removal of all malignant cells. Unfortunately, it is not possible to identify the microscopic extent of tumor invasion by inspection and palpation. To ensure tumor removal, the surgeon excises a “safe” margin around the tumor and uses frozen sections to confirm that the margins of resection do not contain

cancerous tissue. In oral squamous carcinoma, there is lack of agreement as to what constitutes a “safe” margin. The usual consensus is 1 to 2 cm. In 1978, Looser et al,l reviewed their experience with microscopically “positive” margins. They excluded patients with gross residual disease and defined four groups as having microscopically positive margins: patients with cancer within 0.5 cm of the margin, patients with atypia or premalignant changes in the margin, those with carcinoma in situ in the margin, and those with invasive carcinoma in the margin. Interestingly, all four of these groups showed an increase in local recurrence ranging from 64% with invasive carcinoma to 80% with premalignant change. Overall, there was a local recurrence rate of 71% in these groups, which compared to a recurrence rate of 3 1.7% in patients with negative margins. The highest rates of

* Associate Professor, Department of Oral and Maxillofacial Surgery, University of Maryland: Associate Professor, Oncology Pro&a&, University of Ma&land Cancer Center, Baltimore, MD. t Professor Department of Pathologv, UMDNJ-New Jersev Medical School, Newark, NJ. -. Address corrcspondencc and reprint requests to Dr Ord: Department of Oral and Maxillofacial Surgery, University of Maryland at Baltimore, 666 West Baltimore St, Baltimore. MD 21201-1586. 0 1997 American

Association

of Oral and Maxillofacial

Surgeons

0278-2391/97/5507-0002$3.00/O

663

664 recurrence were in patients with stage IV disease; 90% for the positive margin groups and 47.3% for the patients with negative margins. Surprisingly, there was little difference in 5-year survival between the patients with positive and those with negative margins. The authors concluded that failure to achieve negative margins reduced the chance of local control, but that negative margins did not guarantee local control of disease. Byers et al2 examined the use of frozen sections in the determination of positive margins and found that 67% of tumors were adequately excised based on the surgeon’s clinical judgment. A further 10% of tumors could never be adequately excised despite use of frozen sections. However, 23% of cases were correctly identified as having positive margins at the time of excision by the use of frozen sections, which enabled further excision to obtain negative margins. In the oral cavity, local recurrence of cancer was 12% when these were negative margins and 13% in lesions with initially positive margins that were re-excised after frozen section diagnosis to obtain negative margins. This compared to 80% local recurrence in cases with positive margins. These results were mirrored by the 2-year survival, which was 77% and 86% for the two groups with negative margins and 10% for the patients with positive margins.* This study reaffirmed the importance of negative margins and showed that frozen sections help the surgeon avoid the presence of invasive carcinoma or carcinoma in situ at the margin. However, the authors also found that negative margins were unreliable in predicting local control in T4 lesions. There was also no mention in the article of what constituted a safe margin because only involvement of the margin itself by carcinoma or carcinoma in situ was regarded as positive, and no analysis of “close” margins was undertaken. Since the publication of these two papers in 1978, there have been conflicting studies attempting to define what constitutes a positive margin, what the significance of a positive margin is in tumor recurrence, and how much normal tissue should be removed as a “safe” margin to ensure complete tumor removal. Although Chen et al3 found local recurrence rates of 55% in cases with margins microscopically involved by carcinoma, 50% with carcinoma in situ in the margin, 10% with margins less than 0.5 cm from carcinoma, and 100% with gross tumor in the margin, they still found local recurrence of 17% in patients with clear margins. They did not examine the significance of dysplasia. Almost all published work supports the finding of increased local failure rates when the margin is involved by invasive tumor’Z4-6 or when the margin was close to the lesion.5-7 However, not all authors were able to show an increase in local recurrence when the margin was involved by carcinoma in situ.7-9 None of these findings can be applied to laryngeal cancer. Lam et al” found no recurrence with 2-mm margins, and

ACCURACY

OF FROZEN

SECTIONS

IN

ORAL

CANCER

Bauer et al” reported 39 patients with positive margins (15 close, 12 microscopically involved, and 12 grossly involved), only seven (18%), of which showed local recurrence after hemilaryngectomy. Davidson et al” in 1988 used MOHS technique to monitor the extent of mucosal carcinomas in the oral cavity. This technique uses horizontal sectioning monitored by frozen section and is 1,000 times more likely to detect tumor extensions than conventional vertical histologic sections. They found that 70% of oral cancers could be identified microscopically at least 1 cm beyond the margins expected from inspection and palpation. Although thin fingers of tumor consisting of 10 to 20 cells can be accurately traced with this technique, its success depends on in-continuity permeation of the tissues by tumor. If tumor invasion is discontinuous or embolic in nature, MOHS surgery is not able to ensure total eradication of the tumor. Jakobsson et all3 found that 25% of head and neck carcinomas show an embolic pattern of invasion at the primary site. This may be of particular importance in the tongue, where Harrison14 dissected total glossectomy specimens and found separate tumor emboli 1.2 cm beyond the tumor margins. Harrison’s conclusion was that tongue cancers required a 2-cm margin of excision. Unfortunately, this would compel the surgeon to undertake a total glossectomy for all large tumors T2 and above because the average tongue measures 11 X 6.5 cm. These tumor emboli may be “milked” deeply by the action of the tongue musculature. Most oral squamous carcinomas invade laterally rather than vertically, but tongue cancer is frequently deeply invasive. Recent publications have shown the importance of tumor thickness as a bad prognostic factor,” and one report7 has shown that positive margins in deep tissue gave higher rates of recurrence than positive mucosal margins. With mandibular invasion, palpation is not able to assess margins, and frozen sections are not available for bone Macgregor and McDonald’s’6 work showed little lateral spread in the marrow beneath intact mucosa or cortical bone and indicated that a l-cm margin of uninvolved bone would provide a safe margin. Shusterman et alI7 found that only 2% of mandibular resections undertaken for bone invasion showed positive bony margins. They based their margins on clinical examination and panoramic radiographs. The current study examined the accuracy of frozen section interpretation in comparison to the histologic findings on permanent sections when assessing surgical margins for oral squamous cell carcinoma. Also examined was the nature of the positive and negative margins, the role of frozen sections in defining margins, and prognosis with respect to local recurrence. Patients

and Methods

A retrospective survey of 49 consecutive patients with oral cancer treated at the University of Maryland

ORD

Table

AND

665

ABNER

Analysis

1.

Frozen True True False False

negative positive positive negative

of Frozen Section

Permanent

291 13 1 (atypia) 1 (negative) 1 (negative) 301

Total

Section

Table

Accuracy

Section

Patient

292 15 1 (negative) 1 (dysplasia) 1 (invasive cancer) 307

No.

3 1 1 1

by one surgeon (R.A.O.) was undertaken. In all patients, a minimum clearance of 1 cm around the tumor was attempted surgically. All resections were monitored using frozen section guidance . The tissue was taken as samples from all margins of the specimen, anterior, posterior, medial, lateral, and deep. Any area that was regarded as suspicious by the surgeon was also sampled. The entire margin was not examined by frozen sections. The frozen sections were examined by different staff pathologists, as were the final permanent sections of the sametissue block. The frozen and permanent sections were all reviewed by one experienced histopathologist (S.A.) for this study. These sections were compared with the histologic findings from the final complete surgical specimen. Resection margins containing dysplasia, carcinoma in situ, or infiltrating carcinoma, and margins within 5 mm of carcinoma, were regarded as positive. All stageIII and IV patients received postoperative adjuvant radiation therapy, as did stage I and II patients with positive margins. The 49 patients were followed for an average of 3 1.5 months (17 to 45 months), or until they died. Results

The 49 patients were staged using the standard U.I.C.C. TNM system. There were 19 (38.7%) stage IV patients and 26 (53%) of the total group were stage III or IV. Thirty-two patients were male (65%/o),and 16 (50%) of these were stage III or IV. Seventeen patients were female (35%), and 10 (59%) of these were stageIII or IV. Forty patients (81.5%) were white, and 22 (55%) of these were stage III or IV. Only nine patients (18.5%) were black, and four of these (44%), were late stage. A total of 307 frozen sections were

Local Recurrence Margins

Related

to Status

Local

Patients

Stage

l/15 O/8 l/7 5119

I II III IV

39 4 4 2

No.

Margins Negative <5 mm Dysplasia Carcinoma

Analysis Recurrence

(%)

6.6 0 14 26

4.

Current

Status

of Patients

of Outcome

Patient

Recurrence

performed on the 49 patients, an average of 6.2 per patient. The frozen sections were compared with their permanently stained counterparts, and 304 of 307 showed concordance, an accuracy of 99%. There were two false-negative frozen sections and one false-positive section, giving a sensitivity of 86.6% and a specificity of 99.6% (Table 1). Final margins on examination of the whole permanent specimen showed 39 (79.6%) of the patients to have clear margins. Thirtyeight of these patients had negative margins on initial frozen section, whereas one patient had an initial positive margin that required further excision (Table 2). The seven patients who developed local recurrence were analyzed for correlation with the microscopic status of their margins (Table 2). Although 7.7% of patients reported as having clear margins had recurrence locally, 25% of patients with margins that were close or showed dysplasia and 100% of patients with positive margins had recurrence. When the local recurrence group are analyzed for stage of disease,1 of 1.5stageI (6.6%), 0 stage II, one of seven stage III (14%), and five of 19 (26%) stage IV cases had recurrence (Table 3). Thirty-two patients (65%) are alive and free of cancer, eight are dead from that disease(16%), and two are alive with distant metastases(Table 4). In the 32 patients who are alive and free of disease, four (12%) have been treated for second primaries in the upper aerodigestive tract. In the eight patients who died of disease,three had local recurrence only, three had local recurrence and neck nodes, and two had only neck disease.In addition to the six patients with local recurrence, one other patient had recurrence locally that was re-excised. She subsequently died of other causes. Although frozen sections were 99% accurate compared with their permanent counterparts, they were not

Table Table 2. the Final

3.

No. (%)

Follow-up mo. (average)

Recurrence 3 1 1 2

(7.7%) (25%) (25%) (100%)

Alive no disease Dead of disease Dead of other disease Alive with metastases Lost to follow-up

32 8 5 2 2

(65%) (16%) (10.5%) (4%) (4%)

20-47 mo. (32.2 5-26 mo. (13.3 7-37 mo. (19.4 24-37 mo. (30.5

av.) av.) av.) av.)

ACCURACY

Table 5. Correlation Between Positive Margins, and Survival Patient No. Stage Stage Stage Stage

I II III IV

1.5 8 7 19

Positive

Margins (%I 13.3 12.5 28.5 35.7

Tumor

Alive (%I 73.3 87.5 43.0 51.9

Stage,

Dead

of Disease (%I 6 0 28.5 26.3

always helpful in preventing final positive margins. There were 10 patients with a final diagnosis of positive margins. Two of them had invasive tumor at the margin diagnosedon frozen section, but there was skull baseinvolvement, and it was impossible to obtain clear margins. Another patient had dysplasia diagnosed on frozen section, but it also was technically impossible to get clear margins. The remaining seven patients had dysplasia or close margins that were not identified by frozen section, and were only identified on examination of the permanent specimens. One of the patients had a false-negative frozen section resulting from an error in interpretation. Four patients had close margins diagnosed on examination of the surgical specimen. Close margins cannot be diagnosed or prevented by intraoperative frozen section analysis. Two patients had dysplasia at the resection margins of the surgical specimen, that was in areasthat had not been sampled by frozen sections. These were sampling errors. The 10 patients with positive margins included three patients who died of other causes(cardiac) within 1 year of surgery. The other seven patients with positive margins were analyzed for local recurrence (Tables 2, 3). Two patients with infiltrating tumor at the margins and gross residual diseasehad postoperative radiation therapy, but both died of these disease. Three of the five remaining patients (two dysplasias and one close margin) had no radiation and are alive and cancer free. The other two (both with close margins) received radiation therapy and one had local recurrence and died of cancer. One patient had a positive margin identified on frozen section that was re-excised to give final clear margins and remains diseasefree. This was the only patient who benefited from the frozen sections. There were two patients with false-negative frozen sections. One had atypia on final microscopic examination and the cancer recurred. The other patient had a margin positive for invasive carcinoma, but this was not significant becausefive other margins were diagnosed as positive for carcinoma and could not be excised. One patient with false-positive frozen section had further needless excision of normal tissue. As expected, the percentage of positive margins increasedwith the stage of the tumor, although lessthan many other articles have indicated (Table 5). Six of

OF FROZEN

SECTIONS

IN ORAL

CANCER

seven local recurrences were associatedwith stage III and IV disease,which correlated with a lower survival. Discussion In evaluating the significance of surgical margins, there are a number of questions raised by this study and previous publications. How helpful are frozen sections in guiding the surgeon; what is a positive margin, what treatment is indicated for a positive margin; and finally, how confident can the surgeon be that a negative margin ensuresthat all tumor has been removed? Frozen sections have been used both in diagnosis and to evaluate margins. In most publications, accuracy for such diagnosis is 96.8% to 98%.18-20In the head and neck, Remsenet a121reported an accuracy of 96% for their series, which included frozen sections for diagnosisand margins. Interestingly, they found an 8.7% error for the oropharynx. A large series 1,947 frozen sections that evaluated margins in head and neck tumors showed an accuracy of 98.3%, with only 1.9% discrepancy.z2 False-positive results were always less than 1% in these publications. The one false positive in our series occurred in a patient who had a frozen section specimen read as atypia, which was diagnosed as benign on review of permanent sections. This was attributable to an error in histologic diagnosis and resulted in a further excision of 1 cm of normal tissue. Casesthat have previously had radiation therapy are more difficult to diagnose with frozen sectionsbecause of the presence of fibrosis and inflammation. In this series, 38 patients (77.5%) had their tumors adequately excised based only on the surgeon’sjudgment, which compares to 67% in Byers et al’s. series.2 Three patients (6%), had positive margins (two positive for tumor one for dysplasia) that were not removed because of involvement of vital structures, compared with 10% of such casesin Byers et al’s report. However, 23% of Byers et al’s patients had positive margins initially that were excised to give negative margins. Because this reduced the incidence of recurrence, Byers et al concluded that frozen sections were beneficial. Our serieshad only one patient (2%) with intraoperative frozen sections that were initially positive who had further surgery to give final negative margins. This left seven patients (14.5%) with final positive margins who did not benefit from frozen sections. One case had a false-negative benign frozen section diagnosed as benign that was diagnosed as atypia on review of permanent sections, an error of diagnosis. Frozen sections from two other caseswere diagnosed as benign on both frozen and permanent sections, but the patient was found to have dysplasia when the margins of the excised specimen were examined. These were obviously sampling errors. The final four “positive” cases were from tumors with margins within 5 mm of the

ORD

AND

AISNER

tumor. These could not be diagnosed by frozen section because the marginal tissue was normal. There appears to be no way to avoid this problem with frozen section sampling of margins. It is difficult to show that frozen sections were beneficial in this series because only one patient was helped to obtain negative margins. However, there was one false-positive frozen section and seven patients who had tumors with positive margins at final diagnosis that were not diagnosed by the intraoperative frozen sections. There was 100% recurrence when the margins were microscopically involved by infiltrating carcinoma. Patients with close or dysplastic margins showed a 25% recurrence rate, far more than the 7.7% rate seen in patients with negative margins. The one case of carcinoma in situ was included in the dysplasia group because the patient also had a second margin involved with severe dysplasia. Interestingly, this tumor did not recur. The one case with local recurrence in this group was positive for mild atypia only at the margins. The finding of dysplasia at a margin of an excised cancer would logically seem to increase the chances of local failure, although this does not seem to be inevitable. In a recent review of oral dysplasia, 13.8% of 240 lesions became cancers. However, the lesions in 16 of 91 cases (17.5%) improved, regressed; or disappeared after biopsy.23 The behavior of dysplastic tissue is therefore not predictable. There is also a problem with microscopic interpretation of these lesions. General and oral pathologists often use different criteria to grade these cases. Even experienced oral pathologists only agreed 81.5% of the time on distinguishing dysplasia from no dysplasia, and, when “blinded”, only agreed with their own diagnosis of dysplasia versus no dysplasia 80.3% of the time.24 This problem may be increased when trying to identify dysplasia on a frozen section. If close margins are accepted as true positive, then it requires even better intraoperative judgment by the surgeon to obtain a l-cm clearance because frozen sections will never help to diagnose “close margins.” This diagnosis can only be made by careful review of the permanent slides from the excised specimen. Although it appears that margins within 5 mm of tumor are associated with increased recurrence, it is difficult to understand why MOHS surgery should be effective, because this technique only gives microscopic tumor clearance. The reported recurrence after MOHS surgery in the oral cavity was 9% even though postoperative radiation was used in one series.12 Our results indicate that close margins, dysplastic margins, and margins involved by invasive carcinoma should all be regarded as associated with an increased risk of tumor recurrence. We were unable to evaluate carcinoma in situ because there was only one case in our series.

667 The question of what further treatment is required when a positive margin is identified on the final specimen is also debatable. Excision of positive frozen section margins intraoperatively has been shown to give the same recurrence and survival rates as cases with initial negative margins.2 Therefore, immediate excision has been proposed. Leez5 found a 67% 5-year survival for patients who underwent a second excision after inadequate primary surgery, compared with 43% for patients given radiation and 21% for those having no further therapy. However, only 50% of the excised specimens were found to contain tumor; either the frozen section sampling had removed all cancer cells or the excision was not in the correct area. Despite these results, reoperation is difficult because contraction and distortion after reconstruction with soft tissue flaps makes identification of the area of the positive margin difficult. This was shown by Lee’s finding that only 50% of re-excisions contained tumor. With small lesions, it may be possible to leave the defect open until totally clear margins are confirmed, but these are the lesions least likely to have positive margins. As seen in Table 5, only 13% of Tl and T2 tumors had positive margins, and only 1 of 23 patients died of disease (4.3%). In T3 and T4 tumors and composite resections, primary reconstruction with large flaps is required, and accurate re-excision may be technically impossible. Although Lee25 recommended immediate re-excision for all patients with positive margins on the final specimen, his results showed that 5-year survival was also improved with postoperative radiation therapy. In 1980, the classic article from the Memorial Sloan-Kettering Cancer Center showed an 18% recurrence rate in stage III and IV disease for patients who received postoperative radiation therapy versus 50% to 75% in patients treated with surgery alone.26 They also showed that the best results were achieved when radiation was given within 6 weeks of surgery. Despite the fact that this study used historical controls, it has become the “gold standard” for treating advanced head and neck cancer. Because there is an increase in positive margins for T3 and T4 lesions, radiation has been advocated as adjuvant therapy when re-excision is not feasible for residual tumor at the margin. Mantravadi et al4 found that 3-year survival was the same for patients with residual microscopic disease as for those with clear margins after radiation, provided the radiation therapy was given within 6 weeks. Those patients with gross residual disease did not have such a favorable outcome. Other authors have also advocated adjuvant radiation.3 In a careful analysis of multiple factors favoring tumor recurrence, which included stage, tumor thickness, spidery spread, as well as positive margins, Ravasz et al7 concluded that positive margins as a single indicator for use of postoperative radiation was not

668 associated with an increase in recurrence. These results agreed with the findings of Amdur et al.’ Other authors have found little benefit for postoperative radiation in patients with positive surgical marginsz7 In our small series, radiation postoperatively was ineffective for gross residual disease (two of two died of cancer), one of two patients with close margins had recurrence and died of cancer despite adjuvant radiation, and none of three patients with close or dysplastic margins who did not receive radiation had recurrence. These numbers are too small to conclude that radiation is not indicated for patients with positive margins. According to the literature there is a local recurrence rate of 3% to 35% in patients with negative margins. Recurrence is highest in T4 tumors. In Tl and T2 lesions with negative margins and no adjuvant radiation therapy, recurrence rates are as low as 7%.” Mechanisms to explain recurrence in patients with negative margins include errors in histologic diagnosis, sampling errors, and tumors with “embolic” spread. All of these explanations can account for tumor recurrence because the margin was never truly negative. Another possibility is transformation of ‘ ‘normal” tissue at the margin. In 1953, Slaughter et a129proposed their theory of “field cancerization” and discussed the multicentric origin of oral cancer. This concept is widely accepted and has been strengthened by the recognition of second primary tumor development in the upper aerodigestive tract. Recent publications have looked at the molecular events underlying such tumor formation. Discordant ~53 mutations in second primary cancers arising in patients with a primary upper aerodigestive tract cancer give a molecular basis for field cancerization, although the cancers appear to arise as independent events.30 In another series, 83% of head and neck cancer patients with multiple cancers had positive p53 staining in normal aerodigestive mucosa distant from the cancers. This compared with 8.5% positive p53 staining in nonneoplastic controls.3’ An important publication in this field by Brennan et a132 has provided evidence that molecular changes, such as mutant ~53, may be the best marker for local recurrence in patients with “negative” margins. These authors identified 30 oral cancer patients with specific ~53 mutations and, with polymerase chain reaction (PCR) amplification, made unique probes for each tumor. Five patients had grossly or microscopically positive margins (16%) and were excluded from analysis. The other 25 patients with negative margins on light microscopy were analyzed using the molecular probes. Twelve patients were confirmed as negative and none had recurrence locally. Thirteen patients were found to have neoplastic cells containing mutations in the “negative’ ’ margins, and five (38%) had local recurrence. This technique was extremely accurate, identifying 0.05% of abnormal cells in margins previously

ACCURACY

OF

FROZEN

SECTlONS

IN ORAL

CANCER

undiagnosed by light microscopy. The authors concluded that the presence of undiagnosed cells containing ~53 mutations specific for the primary tumor at the margins of resection increases the risk of local recurrence. Although this technique could potentially reduce local recurrence, there are several problems. Only approximately 50% of head and neck tumors show ~53 mutations,32,33 the PCR techniques are expensive and require special equipment and individualized probes, and they cannot be used to identify patients with positive margins at the time of surgery. However, intraoperative molecular analysis of surgical margins may eventually be developed as the “state of the art” method for oncologic surgery. Local recurrence is seen with increasing frequency in tumors with negative margins as size increases. Irrespective of the status of the neck nodes, this would appear to validate the use of adjuvant radiation therapy for all T3 and T4 lesions despite negative margins and for Tl and T2 tumors with positive margins where further excision is not feasible. References 1. Looser KG, Shah JP, Strong EW: The significance of “positive” margins in surgically resected epidetmoid carcinomas. Head Neck 1:107, 1978 2. Byers RM, Bland KI, Borlase B, et al: The prognostic and therapeutic value of frozen section determinations in the surgical treatment of squamous carcinoma of the head and neck. Am J Surg 136:528, 1978 3. Chen TY, Emrich LJ, Driscoll DL: The clinical significance of pathological findings in surgically resected margins of the primary tumor in the head and neck carcinoma. Int J Radiat Oncol Biol Phys 13:833, 1987 4. Mantravadi RVP, Haas RE, Skolnik EM, et al: Postoperative radiotherapy for persistent tumor at the surgical margin in head and neck cancers. Laryngoscope 93:1337, 1983 5. Carter RL. Barr LC. O’Brien CJ. et al: Prognostic imnlications of perineural spread in squamous carcinomas of the’head and neck. Laryngoscope 961145, 1980 6. Lewis LJ, Lavery RS, Groshen S, et al: Postoperative radiation therapy for upper aerodigestive tract cancers with positive surgical margins: prognostic factors. Int J Radiat 0~01 Biol Phys 17:230, 1989 (suppl 1) 7. Ravasz LA, Slootweg PJ, Hordijk GJ, et al: The status of the resection margin as a prognostic factor in the treatment of head and neck carcinoma. J Craniomaxillofac Surg 19:3 14, 1991 8. Amdur RJ, Parsons JT, Mendenhall WM, et al: Postoperative irradiation for squamous cell carcinoma of the head and neck: An analysis of treatment results and complications. Int J Radiat Oncol Biol Phys 16:25, 1989 9. Sadeghi A, Kuisk H, Tran LM, et al: The role of radiation therapy in squamous cell carcinoma of the upper aerodigestive tract with positive margins. Am J Clin Oncol9:500, 1986 10. Lam KH, Lau WF, Wei WI: Tumor clearance at resection margins in total laryngectomy: A clinicopathologic study. Cancer 61:2260, 1988 11. Bauer WC, Lesinski SG, Ogura JH: The significance of positive margins in hemilaryngectomy specimens. Laryngoscope 85:1, 1975 12. Davidson TM, Haghighi P, Baird P, et al: MOHS for head and neck mucosal cancer: Report on 111 patients. Laryngoscope 98:1078, 1988 13. Jakobsson PA, Eneroth GM, Killander D, et al: Histologic classi-

WILLIAM

14. 15.

16. 17.

18.

19.

20.

21.

22.

23.

669

.I. FRABLE

fication and grading of malignancy in carcinoma of the larynx. Acta Radio1 12:1, 1973 Harrison DFN: The questionable value of total glossectomy. Head Neck Surg 6:632, 1983 Spiro RH, Huvos AG, Wong JD, et al: Predictive value of tumor thickness in squamous carcinoma confined to the tongue and floor of the mouth. Am J Surg 152:345, 1986 McGregor AD, McDonald DG: Patterns of spread of squamous cell carcinoma within the mandible. Head Neck 11:457, 1989 Shusterman MA, Harris SW, Raymond AK, et al: Immediate free flap mandibular reconstnrction: Significance of adequate surgical margins. Head Neck 15:204, 1993 Kaufman Z, Lew S, Griffel B, et al: Frozen-section diagnosis in surgical pathology: A prospective analysis of 526 frozen sections. Cancer 57:377, 1986 Rogers C, Klatt EC, Chandrasoma P: Accuracy of frozen section diagnosis in a teaching hospital. Arch Path01 Lab Med 111:514, 1987 Howanitz PJ, Hoffman GG, Zarbo RJ: The accuracy of frozen section diagnoses in 34 hospitals. Arch Path01 Lab Med 114:355, 1990 Remsen KA, Lucente FE, Biller HF: Reliability of frozen section diagnosis in head and neck neoplasms. Larygoscope 94:519, 1984 Gandour-Edwards RF, Donald PJ, Wiese DA: Accuracy of intraoperative frozen section diagnosis in head and neck surgery: Experience at a university medical center. Head Neck 15:33, 1993 Lumerman H, Freedman P, Kerpel S: Oral epithelial dysplasia and the development of invasive carcinoma. Oral Surg Oral Med Oral Path01 79:321, 1995

24. Abbey LM, Kaugars GE, Gunsolley JC, et al: Intraexaminer and interexaminer reliability in the diagnosis of oral epithelial dysplasia. Oral Surg Oral Med Oral Path01 80:1X8, 1995 25. Lee JG: Detection of residual carcinoma of the oral cavity, oropharynx, hypopharynx, and larynx: A study of surgical margins. Trans Am Acad Ophthalmol Otol 78:49, 1974 26. Vi&ram B, Strong EW, Shah J, et al: Elective postoperative radiation therapy in stages III and IV epidermoid carcinoma of the head and neck. Am .I Surg 140:5X0, 1980 27. Zieske LA, Johnson JT, Myers EN, et al: Squamous cell carcinoma with positive margins: Surgery and postoperative radiation. Arch Otolaryngol Head Neck Surg 112:863, 1986 28. Ikemura K, Ohya R: The accuracy and usefulness of frozen section diagnosis. Head Neck 12:298, 1990 29. Slaughter DP, Southwick HW, Smejkal W: “Field cancerization” in oral stratified squamous epithelium: Clinical implications of multicentric origin. Cancer 6:963, 1953 30. Chung KC, Mukhopadhyay T, Jhingook K, et al: Discordant p53 gene mutations in primary head and neck cancers and corresponding second primary cancers of the upper aero digestive tract. Cancer Res 53:1676, 1993 3 1. Gallo 0, Bianchi S: p53 expression: A potential biomarker for risk of multiple malignancies in the upper aerodigestive tract. Oral Oncol Eur J Cancer 31B:53, 1995 32. Brennan JA, Mao L, Hruban RH, et al: Molecular assessment of histopathological staging in squamous cell carcinoma of the head and neck. N Engl J Med 332:429, 1995 33. Boyle J, Hakim J, Koch W, et al: The incidence of p53 mutations increases with progression of head and neck cancer. Cancer Res 5314477, 1939

J Oral Maxillofac Surg 55:669-671, 1997

Discussion Accuracy of Frozen Sections Margins in Oral Cancer

in Assessing Resection

William J. Frable, MD Virginia Commonwealth Richmond, Virginia

University,

Medical

College

of Virginia,

In a series of 307 frozen sections from 49 patients used to assess margins at the time of resection of oral carcinomas, the authors found an accuracy of 99%. There were two falsenegative reports and one false-positive report when the frozen section diagnosis was compared with the final diagnosis on permanent sections. This confirms the high accuracy of this procedure, as the authors also document from the literature. More important is the effect on treatment outcome. This question is particularly pertinent in today’s environment of health care cost containment. Although this series is small, it is the experience of a single surgeon performing all of the resections. The authors conclude that only one patient benefited from an intraoperative re-excision of a positive margin to give a negative margin. One other patient had a false-positive frozen section and had additional unnecessary surgery. A general conclusion from the literature review by the authors is that assessment of the margins predicts local recurrence if they are positive or close (5.0 mm or less) The several series reviewed all have some notable deficiencies

or lack important details. In the series by Looser et al,’ there was no impact on survival between patients with negative and positive margins only a difference in local control. Seventeen patients received variable doses preoperative radiation therapy to the neck, or the neck and the primary, a finding not controlled for in that study. Five margins were examined, but it does not say if this was true in all cases, and there is no indication of which margin was involvedclosest, deep, anterior, posterior, lateral, or medial. Fortyone of the patients in this series were stage III or IV, patients whose prognosis is not very good based on staging. The risk for failure locally was the same whether the margin was gross tumor, premalignant changes, carcinoma in situ or close, 0.5 mm. Many of the failures may represent new primary tumors in a mucosa already programmed to develop carcinoma. This is often confirmed histologically when biopsies from recurrences show associated carcinoma in situ. The conclusions from the study by Looser et al’ with respect to pathologic examination were not supported by the data presented: all surgical specimens should be routinely processed to study “all margins” of resections; after removal of the specimen, margins of the surgical defect should be checked in all cases to ensure adequacy of resection. What does “all margins” mean in this context? From the perspective of surgical pathology, “all margins” is impractical and intraoperatively would increase operating time to an unacceptable length. Other studies reviewed by the Ord and Aisner report some