- Email: [email protected]
Do frozen sections help achieve adequate surgical margins in the resection of oral carcinoma?
Int. J. Oral Maxillofac. Surg. 2003; 32: 152–158 doi:10.1054/ijom.2002.0262, available online at http://www.sciencedirect.com
Clinical Paper Head and Neck Oncology
Do frozen sections help achieve adequate surgical margins in the resection of oral carcinoma?
N. F. F. Ribeiro1, D. R. P. Godden1, G. E. Wilson2, D. M. Butterworth2, R. T. M. Woodwards1 1
Department of Oral and Maxillofacial Surgery and 2Histopathology, North Manchester General Hospital, Manchester, UK
N. F. F. Ribeiro, D. R. P. Godden, G. E. Wilson, D. M. Butterworth, R. T. M. Woodwards: Do frozen sections help achieve adequate surgical margins in the resection of oral carcinoma? Int. J. Oral Maxillofac. Surg. 2003; 32: 152–158. 2003 International Association of Oral and Maxillofacial Surgeons. Published by Elesevier Science Ltd. All rights reserved. Abstract. This retrospective study examined the notes of 82 patients who underwent an attempted curative resection of oro-pharyngeal carcinoma, and who had frozen sections taken, over a four year period. Three hundred and fifty mucosal, 179 deep tissue, and 22 nerve frozen sections had been taken. Concordance between cryostat and paraffin sections was 99.5% although no false positives or negatives for invasive tumour were found. Nine (11%) patients who had frozen sections which were reported as dysplastic or positive for invasive tumour underwent further local resection; excision was then found to be complete in 8 (10%) of these patients with further frozen sections. In 15 patients the margins of the main resection specimen were judged to be close to the tumour (<5 mm), 5 patients had squamous epithelial dysplasia and 12 patients had invasive tumour at a resection margin. Ten of the 12 patients with margins containing invasive tumour had negative cryostat sections intraoperatively, which demonstrated problems with sampling which is the major drawback. During follow up, 14 patients developed local recurrence. Only one of the patients who had frozen section guided further local resection, developed local recurrence. Frozen sections help achieve tumour clearance.
Introduction Resection margins that are clear of tumour are of paramount importance in optimizing treatment of malignant oropharyngeal tumours. Failure to achieve clear margins reduces the chance of local control2,3,12,22. Recurrence most commonly occurs when the generally accepted 1 cm margin of normal tissue is not achieved around the tumour at the time of resection. An 0901-5027/03/020152+07 $30.00/0
involved margin is defined as invasive tumour, carcinoma in situ or squamous epithelial dysplasia at the resection margin, or a margin judged to be within 5 mm of the tumour12. Studies reveal that between 9% and 47% of cases have positive margins, Table 12–4,9,12,13,15,20. To help overcome this problem, in some units, frozen section analysis is undertaken from the mucosal and deep surfaces of the defect intraoperatively; if tumour remains then further resection
Key words: frozen sections; tumour resection; resection margins; oral cancer. Accepted for publication 11 February 2002
can be undertaken at the time of the primary resection. The value of frozen section analysis is, however, controversial and its usefulness has been questioned19, as resection margins have shown a low yield of tumour after an initial specimen was reported positive17. The aims of this study were to investigate the usefulness of frozen sections in head and neck oncological resections, and to determine if their routine use was
2003 International Association of Oral and Maxillofacial Surgeons. Published by Elsevier Science Ltd. All rights reserved.
153
Frozen sections analysis Table 1. Involved margins in other studies
Table 2. Stage of tumour
Percentage
Authors 22
8.8% (Invasive tumour only) 13% (Invasive tumour only) 14% (Invasive tumour only) 55% T4 21% T1 (Invasive tumour only) 12.9% 20% 39% 47% (Excluding dysplasia)
justified. It examined the accuracy of frozen sections in defining margins and determined whether frozen sections improve the rate of tumour clearance, which is the goal in surgical oncology. The study also analysed the reasons why there may be failure to achieve surgical clearance. Method A retrospective analysis was performed on the clinical notes of patients who had undergone an attempted curative tumour resection for oropharyngeal carcinoma, and who had frozen sections taken, between August 1994 and October 1998. Eighty-two patients were identified; all were treated by one of the authors (RTMW). These patients were followed up and their outcome as of the first of August 2001 recorded. It is the normal practice of the surgeon to undertake frozen sections from the mucosal and deep margins of the resection. If these are found to be positive for squamous epithelial dysplasia or invasive tumour, then further local resection is performed if practicable. The main specimen was subsequently examined as was tissue taken when frozen sections dictated further local resection for clearance. When bone was also resected decalcified sections were analysed. Resection margins containing squamous epithelial dysplasia, carcinoma in situ, infiltrating carcinoma and margins within 5 mm of carcinoma were considered positive in our study in keeping with the definition of involved margins proposed by L et al.12. Patients with frank tumour at the margins and those who had positive nodes in the neck with extracapsular spread were considered for postoperative radiotherapy. Biopsy tissue submitted for frozen section analysis was frozen with a freezing mixture of isopentene and solid carbon dioxide (Cardice). Sections of 5–7 m were then made, adhered to a glass slide at room temperature, fixed with formol
Z et al. C et al.4 J et al.9 L et al.13 C et al.3 O et al.15 W et al.21 W et al.20
acetic alcohol (50 ml formalin, 450 ml 90% alcohol and 25 ml glacial acetic acid) and stained with haematoxylin and eosin. After the sections were deemed satisfactory the remainder of the tissue was stored in 10% buffered formaldehyde and routinely processed; embedded in paraffin, sectioned to 3 m and stained with haematoxylin and eosin. The study examined the sensitivity and specificity of frozen sections, when compared to paraffin preparations of the same sample. It also examined whether tissue resected from around the surgical defect, dictated by frozen sections analysis, was involved. The nature of the final surgical margins on the main specimen and the clinical outcome of the patients were also evaluated. Results Fifty-eight male and 24 female patients were reviewed, with a male to female ratio of 5:2. The mean age, at the time of surgery, for males was 63 years (range: 36 years to 84 years); and that for females was 69 years (range: 38 to 89 years). The stage of the tumour is shown in Table 2, and site of tumour is shown in Table 3. Three hundred and fifty mucosal, 179 deep tissue and 22 nerve sections were taken. There was concordance between the cryostat and subsequent paraffin preparation in 548 of 551 biopsies (99.5%). There were no false positives or false negatives for invasive tumour when frozen sections were compared with subsequent paraffin sections. One patient had a cryostat section which indicated the presence of squamous epithelial dysplasia but which was later reported to be normal tissue and two patients had negative frozen sections that were later reported as showing squamous epithelial dysplasia on the paraffin sections. The Sensitivity was 92.8%, the Specificity 99.8% and the Kappa value was calculated to be 0.943 (Kappa values between 0.81–1.00 show almost perfect agreement).
Stage of tumour
Number of patients (n=82)
I II III IV
14 15 19 34
(17%) (18%) (23%) (42%)
Table 3. Site of tumour
Site of tumour Tongue Floor of mouth Lower alveolus and mandible Buccal mucosa Palate Retromolar area Lip Other
Number of patients (n=82) 25 23 8 6 5 5 3 7
Nine patients who had frozen sections reported as containing dysplastic squamous epithelium or invasive tumour underwent further local resection (Table 4); excision was then judged to be complete in eight of these patients with the help of further frozen sections. Tissue that was resected because of positive frozen sections was found to be positive for invasive tumour in three patients and for dysplasia in one patient. This tissue was found to be negative in four of these cases. In one patient (case 8, Table 4) it appears the tissue was inadvertently processed with the main specimen and this has therefore not been reported. In case 2 (Table 4) although two deep frozen sections were positive for tumour intraoperatively and tissue subsequently resected was positive for tumour, the deep final resection margins appeared free from tumour as far as could be assessed. There was extensive vascular permeation by the tumour in the specimen and lymph nodes in the specimen were positive with extracapsular spread. Similarly in case 5 (Table 4) further local resection was carried out because frozen section identified tumour confirmed to be an incidental neurofibroma later in paraffin preparation. Further resected tissue, however, was positive with a focus of carcinoma. Finally, one patient (case 3, Table 4) had an extensive T4 lesion in the maxilla with skull base involvement where tumour clearance was not possible. There were eight patients who had mild dysplasia on frozen sections intraoperatively but in whom further local resection was not carried out. Seven of
154 Ribeiro et al.
Table 4. Patients who had further local resection dictated by frozen sections
Patient
pTNM
Frozen section
Further local resection tissue
Main specimen margins Mucosal
Deep
Bone
Further management
1 2
T4N1M0 T4N1M0
2 Carcinoma in situ 2 Deep Frank tumour
Dysplasia Frank tumour
Dysplasia 4 mm
2 mm 7 mm
Clear —
Radiotherapy Radiotherapy
3 4
T4N2M0 T1N0M0
Frank tumour Negative
Frank tumour 10 mm
Frank tumour 10 mm
Frank tumour —
Radiotheraphy —
5
T2N2M0
7 mm
—
T4N0M0 T2N0M0
Ca in tissue, dysplasia at the margin Negative Negative
8 mm
6 7
1 Frank tumour 2 FS Dysplasia PS inflammation only FS reported positive found to be incidental neurofibroma on PS FS ?Reactive changes 3 Dysplasia
6 mm 5 mm
4 mm 7 mm
Clear —
Pt declined radiotheraphy Radiotherapy Radiotherapy
8 9
T4N2M0 T1N0M0
1 Frank tumour 1FS Dysplasia 2PS Dysplasia
— Negative
4 mm 7 mm
Frank tumour 8 mm
Clear —
Radiotherapy —
NSR—no sign of local recurrence, NCL—no palpable cervical lymphadenopathy, FS—frozen section, PS—paraffin section.
Local recurrence/ outcome NSR NCL NSR died of neck disease Yes died of disease NSR NCL NSR died of neck disease NSR NCL NSR died of neck disease NSR NCL NSR NCL Died of other disease
Frozen sections analysis
155
Table 5. Main specimen margins and outcome Frozen sections Margins
T Stage
Recurrence
No. of patients
Dysplastic
Ca in situ
Frank tumour
T1
T2
T3
T4
Local
Neck
Died of disease
50
6
—
—
11
22
5
12
2
14
13
7 8 5
1 — 2
— — 1
1 — —
1 — 1
— 3 2
1 2 1
5 3 1
— 2 2
2 2 1
1 2 1
3 7 1 1
— — 1 —
— — — —
— 1 — 1
1 — — —
— — — —
— 2 — —
2 6 1 1
3 3 1 1
1 2 1 1
2 3 1 1
Clear margins Close margins (<5 mm) Mucosal Deep Dysplasia/Ca in situ in mucosal margins Frank tumour at margin Mucosal Deep Bone Mucosal deep and bone
Fig. 1. Shows a close margin (<5 mm): invasive, rather poorly differentiated tumour extending to within about 1 mm (A) of the resection margin (B). H&E 50.
these patients had T1–T2 lesions where a wide local resection was performed and further local resection was considered impracticable. Five of these patients had clear final resection margins, two had dysplastic margins and one patient had bone involvement in the final surgical specimen. As shown in Table 5, in 15 (18%) patients the margins of the subsequent main resection specimen were judged to be close to the invasive tumour (<5 mm), Fig. 1, five (6%) patients had dysplasia and 12 (15%) patients had tumour at a resection margin. Of the 12 patients with frank tumour at a resection margin, seven had deep margins involved on the main specimen. Three of these patients had adjacent bone removed, which was clear and might
have represented the true resection margin. The patients were followed up for between 83 and 33 months postoperatively, mean: 55 months (SD 13.90), or until they died. There were 14 (17%) patients who developed local recurrence between 2 months to 46 months (median: 6.98 months) postoperatively. Of these, 11 (79%) had stage IV disease. Two (14%) patients had close (<5 mm) deep margins on the main surgical specimen and eight (57%) patients had invasive tumour at a resection margin. Of the eight patients with invasive tumour at the main specimen margin three had mucosal, three had deep tissue, one had bone invaded and the final patient, in whom clearance could not be achieved, had mucosal, deep and bone margins invaded by tumour. Four (29%) patients who developed local recurrence had resection margins that were negative. To date, 24 (29%) patients have died of their disease, of these five (21%) died as a result of recurrence at the primary site only, six (25%) died with both local and cervical/metastatic disease and 13 (54%) died of cervical/metastatic disease with no local recurrence. The median survival time for the patients who died was 11.78 months. In addition, 17 patients died from other causes.
Discussion The complete excision of tumour is an unquestionable principle of surgical oncology. The majority of patients that present to our unit, that have been represented in this study, have extensive disease at presentation (42% presented with stage IV disease in this study) and often have significant co-morbidity reflected by the proportion of patients
(21%) that died from other causes. W et al.21 have shown that survival is influenced by stage of disease, pattern of invasion and by the presence of tumour at or near the resection margin. Frozen sections may help improve the outcome of patients with margins involved by tumour. L11, in his study of 1522 cases of aerodigestive tract carcinoma, has shown that the oral cavity was the most likely site to have margins involved with residual tumour, a rate of 15% for margins positive for tumour, a figure similar to that found in this study. He recommended intraoperative control of margins with frozen sections and prompt re-excision for an optimal result, although only 50% of re-excisions were found to contain tumour. In our study nine (11%) patients had a further local resection based on the finding of invasive tumour or squamous epithelial dysplasia on frozen section analysis. Eight of these patients were subsequently judged to be clear of invasive tumour on further frozen sections taken at the time of surgery. The tissue subsequently resected was only evaluated for positivity—its margins could not be assessed because of problems with orientation and interpretation, particularly where deep tissue was resected. In four of eight cases i.e. 50% where this tissue was examined there was tumour involvement or dysplasia, a figure similar to that reported by L11 for re-excisions undertaken when resection margins were reported positive for tumour. Almost all these patients, Table 4, received postoperative radiotherapy and only the patient in whom clearance could not be achieved at the time of primary resection developed local recurrence. Three of these patients, however, died of neck disease. L et al.12 have shown a high rate of recurrence (74%) in close margins
156
Ribeiro et al.
Fig. 2. No epithelium identified on the frozen section. H&E 200.
Fig. 3. Paraffin preparation of the same sample shown in Fig. 2, which shows mild to moderate dysplasia. H&E 200.
<5 mm from tumour. Of the 82 patients in our study two (13%) of 15 patients with close margins developed local recurrence as opposed to 2 (4%) of 50 patients with clear margins. Frozen sections do not help improve close margins. Also allowances are not usually made for tissue shrinkage when the margins are judged by the surgeon. This may be interpreted as close margins, after tissue shrinkage, by the histopathologist. J et al.8 in their study on mongrel dogs have shown that the mean shrinkage from initial resection to final microscopic assessment of the lingual surface mucosal margins was 30.7% (3.7 mm), the deep tongue margin 34.5% (5.4 mm) and that for labiobuccal mucosal margins was 47.3% (5.7 mm). The greatest proportion of shrinkage occurred immediately upon resection, there being no disproportionate shrinkage observed. They recommend that an in situ margin of resection of at least 8 to 10 mm need to be taken to obtain 5 mm of pathologically clear margins.
Eight patients in our study had dysplastic frozen sections but further resection was not considered appropriate. Squamous epithelial dysplasia is often extensive with wide field changes often seen in patients with oropharyngeal carcinomas, making adequate excision impracticable in this group of patients. One of the purposes of this study was to assess whether there was failure to identify dysplastic epithelium or invasive carcinoma in frozen tissue. Histological examination of frozen tissue is now commonplace within surgical pathology practice in many units and plays an important role in decisions made by the surgeon during an operative procedure. The procedure is however time consuming, costly and sometimes stressful. It is also well known by surgeons and histopathologists that there are sampling errors and interpretational difficulties in the assessment of frozen tissue in which cellular morphology is not as clearly defined as in formalin fixed and then routinely processed tissue.
In this study resection margin tissue biopsies were examined particularly for the presence of epithelial dysplasia and invasive tumour. A potential difficulty in such biopsies is in differentiating between the reactive epithelial changes seen as a consequence of mucosal inflammation and true epithelial dysplasia. Another potential difficulty is that only a small number of sections of such biopsies can be taken due to the necessary constraints of time and it is therefore possible that dysplastic or neoplastic changes may not be sampled since they are present deeper within the frozen tissue block. This is a well recognized limitation of frozen section examination. Neither of these potential problems proved significant when the findings of frozen section were compared with those of subsequent paraffin embedded tissue. There was concordance in 548 of 551 biopsies, resulting in an overall accuracy of over 99.5%. Accuracy rates of between 96% and 99% have been previously reported6,7,11,17. Processing error occurred in one of the cases with no concordance between the cryostat and subsequent paraffin section (Figs 2 and 3) here the epithelium was not present in the cryostat section, missing dysplasia which was seen in the subsequent paraffin section. In the second case, although the frozen section was negative, a small area of dysplasia was detected in the subsequent paraffin section; a dysplastic region was therefore missed, as it was present in another region of the tissue. In the third case mucosal inflammation gave the impression of dysplasia in the frozen section, which was subsequently reported as normal tissue in the paraffin preparation (Figs 4 and 5). Despite the very satisfactory performance of frozen section reporting in this study, there has been failure to achieve adequate clearance of invasive tumour in 12 (15%) patients. Ten of these 12 patients who had invasive tumour at the final resection margin had clear frozen sections intraoperatively. Clearly the main issue is that the surgeon can take only a small number of biopsies from the resection margin, which represent only a very small percentage of the resection margin. Seven of these patients had invasive tumour at the deep margins. The deep margins are particularly difficult to assess intraoperatively and additional frozen sections taken from these margins might facilitate clearance in selected cases.
Frozen sections analysis
157
Frozen sections are accurate, however their shortfall is susceptibility to sampling errors. They are only as accurate as the specimen taken.
References
Fig. 4. Some epithelial cells have apparently enlarged nuclei, possibly also hyperkeratotic. This was therefore reported as showing mild to moderate dysplasia. H&E 200.
Fig. 5. Paraffin preparation of the same sample as shown in Fig. 4 shows epithelium that looks more orderly with chronic inflammatory cells producing reactive epithelial changes. There was no dysplasia. H&E 200.
Improved control has also been demonstrated following postoperative radiotherapy when there has been a positive surgical margin14,18,22. V et al.18 have shown that 10.5% of patients with positive margins treated with postoperative radiotherapy developed recurrence compared with 73% of patients who did not receive radiotherapy. Patients with tumour at a resection margin subsequently underwent radiotherapy; of these patients 8 developed local recurrence, and 7 patients died from their disease. Our findings concur with others with respect to subsequent local recurrence i.e. high incidence of recurrence in patients with positive margins on the final resection specimen (86%). However 2 of 50 patients (14%) with negative surgical margins on the main resection specimen developed local recurrence. Tumour recurrence can still occur
despite histologically clear margins. This may be because islands of tumour cells remain which cannot be detected microscopically. Brennon et al.1 have shown in a study using molecular techniques that clonal populations of infiltrating tumour cells harbouring mutations of the p53 were detected in 52% of specimens of their patients with histopathologically negative margins, and 38% of patients who had molecular evidence of such mutations at a surgical margin developed recurrence, compared with none in the negative margin group. Molecular biology, however, is not available perioperatively. The development of a more sensitive tool to ensure complete tumour excision is awaited. Frozen sections improve the prospect of the resection achieving tumour clearance and their routine use is justified. They do not influence the outcome of patients with close resection margins.
1. B AJ, M L, H RH, B JO, E YJ, K MW, G SN, S D. Molecular assessment of histopathological staging in squamous cell carcinoma of the head and neck. N Engl J Med 1995: 332: 429–435. 2. B RM, B KI, B B, L M. The prognostic and therapeutic value of frozen section determinations in the surgical treatment of squamous carcinoma of the head and neck. Am J Surg 1978: 136: 528. 3. C TY, E LJ, D DL. The clinical significance of pathological findings in surgically resected margins of the primary tumour in head and neck carcinoma. Int J Radiat Oncol Biol Phys 1987: 13: 833–837. 4. C JA, J AS, P DE, S L E. Implications of tumour in resection margins following surgical treatment of squamous cell carcinoma of the head and neck. Clin Otolaryngol 1993: 18: 37–41. 5. F MJ, L JD. The changing pattern of oral cancer 1977–1995. Br J Oral Maxillofac Surg 1995: 33: 328. 6. G-E R, D PJ, W D. The accuracy and clinical utility of frozen section diagnosis in head and neck surgery. Experience at a university medical centre. Head Neck 1993: 15: 33–38. 7. H PJ, H GG, Z RJ. The accuracy of frozen section diagnoses in 34 hospitals. Arch Pathol Lab Med 1990: 114: 355. 8. J RE, S JD, F GF, R RA, H HT. Quantification of surgical margin shrinkage in the oral cavity. Head Neck 1997: 19(4): 281– 286. 9. J AS, B H Z, N V, R NJ, K A, H TR. Do positive resection margins after ablative surgery for head and neck cancer adversely affect prognosis? A study of 352 patients with recurrent carcinoma following radiotherapy treated by salvage surgery. Br J Cancer 1996: 74: 128– 132. 10. K Z, L S, G B, D A. Frozen-section diagnosis in surgical pathology: a prospective analysis of 526 frozen sections. Cancer 1986: 57: 377. 11. L JG. Detection of residual carcinoma of the oral cavity, oropharynx, hypopharynx and larynx: a study of surgical margins. Trans Am Acad Ophthalmol Otolaryngol 1974: 78: 49–53.
158
Ribeiro et al.
12. L KG, S JP, S EW. The significance of ‘‘positive’’ margins in surgically resected epidermoid carcinomas. Head Neck Surg 1978: 1: 107–111. 13. L TR, S EW. Significance of positive margins in oral cavity squamous carcinoma. Am J Surg 1990: 160: 110–114. 14. M RVP, H RE, L EJ, et al. Postoperative radiotherapy for persistent tumour at the surgical margin in head and neck cancers. Laryngoscope 1983: 93: 1037–1040. 15. O RA, A S. Accuracy of frozen sections in assessing margins in oral cancer resection. J Oral Maxillofac Surg 1997: 55: 663–669. 16. R C, K EC, C P. Accuracy of frozen section diagnosis in a teaching hospital. Arch Pathol Lab Med 1987: 111: 564.
17. S P, B RM, B JG, W P, S H. Microscopic cutthrough of cancer in the surgical treatment of squamous carcinoma of the tongue. Am J Surg 1986: 152: 354–360. 18. V B, S EW, S JP, S R. Failure at the primary site following multimodality treatment in advanced head and neck cancer. Head Neck Surg 1984: 6: 720–723. 19. W JA. T2 carcinoma of the tongue: the histopathologists perspective. Br J Oral Maxillofac Surg 1999: 37: 187– 193. 20. W JA, R S, W CR, E ER, B JS, V ED. Survival and patterns of recurrence in 200 oral cancer patients treated by radical surgery and neck dissection. Oral Oncol 1999: 35: 257–265.
21. W JA, S J, V ED, B JS, W CR, R S. Survival, metastasis and recurrence of oral cancer in relation to pathological features. Ann R Coll Surg Engl 1995: 77: 325–331. 22. Z AL, J JT, M EN, T PB. Squamous cell carcinoma with positive margins. Arch Otolaryngol Head Neck Surg 1986: 112: 863–866. Address: Mr Noel F. F. Ribeiro Maxillofacial Unit North Manchester General Hospital Delaunays Road Crumpsall Manchester M8 5RB UK
Clinical Paper Head and Neck Oncology
Do frozen sections help achieve adequate surgical margins in the resection of oral carcinoma?
N. F. F. Ribeiro1, D. R. P. Godden1, G. E. Wilson2, D. M. Butterworth2, R. T. M. Woodwards1 1
Department of Oral and Maxillofacial Surgery and 2Histopathology, North Manchester General Hospital, Manchester, UK
N. F. F. Ribeiro, D. R. P. Godden, G. E. Wilson, D. M. Butterworth, R. T. M. Woodwards: Do frozen sections help achieve adequate surgical margins in the resection of oral carcinoma? Int. J. Oral Maxillofac. Surg. 2003; 32: 152–158. 2003 International Association of Oral and Maxillofacial Surgeons. Published by Elesevier Science Ltd. All rights reserved. Abstract. This retrospective study examined the notes of 82 patients who underwent an attempted curative resection of oro-pharyngeal carcinoma, and who had frozen sections taken, over a four year period. Three hundred and fifty mucosal, 179 deep tissue, and 22 nerve frozen sections had been taken. Concordance between cryostat and paraffin sections was 99.5% although no false positives or negatives for invasive tumour were found. Nine (11%) patients who had frozen sections which were reported as dysplastic or positive for invasive tumour underwent further local resection; excision was then found to be complete in 8 (10%) of these patients with further frozen sections. In 15 patients the margins of the main resection specimen were judged to be close to the tumour (<5 mm), 5 patients had squamous epithelial dysplasia and 12 patients had invasive tumour at a resection margin. Ten of the 12 patients with margins containing invasive tumour had negative cryostat sections intraoperatively, which demonstrated problems with sampling which is the major drawback. During follow up, 14 patients developed local recurrence. Only one of the patients who had frozen section guided further local resection, developed local recurrence. Frozen sections help achieve tumour clearance.
Introduction Resection margins that are clear of tumour are of paramount importance in optimizing treatment of malignant oropharyngeal tumours. Failure to achieve clear margins reduces the chance of local control2,3,12,22. Recurrence most commonly occurs when the generally accepted 1 cm margin of normal tissue is not achieved around the tumour at the time of resection. An 0901-5027/03/020152+07 $30.00/0
involved margin is defined as invasive tumour, carcinoma in situ or squamous epithelial dysplasia at the resection margin, or a margin judged to be within 5 mm of the tumour12. Studies reveal that between 9% and 47% of cases have positive margins, Table 12–4,9,12,13,15,20. To help overcome this problem, in some units, frozen section analysis is undertaken from the mucosal and deep surfaces of the defect intraoperatively; if tumour remains then further resection
Key words: frozen sections; tumour resection; resection margins; oral cancer. Accepted for publication 11 February 2002
can be undertaken at the time of the primary resection. The value of frozen section analysis is, however, controversial and its usefulness has been questioned19, as resection margins have shown a low yield of tumour after an initial specimen was reported positive17. The aims of this study were to investigate the usefulness of frozen sections in head and neck oncological resections, and to determine if their routine use was
2003 International Association of Oral and Maxillofacial Surgeons. Published by Elsevier Science Ltd. All rights reserved.
153
Frozen sections analysis Table 1. Involved margins in other studies
Table 2. Stage of tumour
Percentage
Authors 22
8.8% (Invasive tumour only) 13% (Invasive tumour only) 14% (Invasive tumour only) 55% T4 21% T1 (Invasive tumour only) 12.9% 20% 39% 47% (Excluding dysplasia)
justified. It examined the accuracy of frozen sections in defining margins and determined whether frozen sections improve the rate of tumour clearance, which is the goal in surgical oncology. The study also analysed the reasons why there may be failure to achieve surgical clearance. Method A retrospective analysis was performed on the clinical notes of patients who had undergone an attempted curative tumour resection for oropharyngeal carcinoma, and who had frozen sections taken, between August 1994 and October 1998. Eighty-two patients were identified; all were treated by one of the authors (RTMW). These patients were followed up and their outcome as of the first of August 2001 recorded. It is the normal practice of the surgeon to undertake frozen sections from the mucosal and deep margins of the resection. If these are found to be positive for squamous epithelial dysplasia or invasive tumour, then further local resection is performed if practicable. The main specimen was subsequently examined as was tissue taken when frozen sections dictated further local resection for clearance. When bone was also resected decalcified sections were analysed. Resection margins containing squamous epithelial dysplasia, carcinoma in situ, infiltrating carcinoma and margins within 5 mm of carcinoma were considered positive in our study in keeping with the definition of involved margins proposed by L et al.12. Patients with frank tumour at the margins and those who had positive nodes in the neck with extracapsular spread were considered for postoperative radiotherapy. Biopsy tissue submitted for frozen section analysis was frozen with a freezing mixture of isopentene and solid carbon dioxide (Cardice). Sections of 5–7 m were then made, adhered to a glass slide at room temperature, fixed with formol
Z et al. C et al.4 J et al.9 L et al.13 C et al.3 O et al.15 W et al.21 W et al.20
acetic alcohol (50 ml formalin, 450 ml 90% alcohol and 25 ml glacial acetic acid) and stained with haematoxylin and eosin. After the sections were deemed satisfactory the remainder of the tissue was stored in 10% buffered formaldehyde and routinely processed; embedded in paraffin, sectioned to 3 m and stained with haematoxylin and eosin. The study examined the sensitivity and specificity of frozen sections, when compared to paraffin preparations of the same sample. It also examined whether tissue resected from around the surgical defect, dictated by frozen sections analysis, was involved. The nature of the final surgical margins on the main specimen and the clinical outcome of the patients were also evaluated. Results Fifty-eight male and 24 female patients were reviewed, with a male to female ratio of 5:2. The mean age, at the time of surgery, for males was 63 years (range: 36 years to 84 years); and that for females was 69 years (range: 38 to 89 years). The stage of the tumour is shown in Table 2, and site of tumour is shown in Table 3. Three hundred and fifty mucosal, 179 deep tissue and 22 nerve sections were taken. There was concordance between the cryostat and subsequent paraffin preparation in 548 of 551 biopsies (99.5%). There were no false positives or false negatives for invasive tumour when frozen sections were compared with subsequent paraffin sections. One patient had a cryostat section which indicated the presence of squamous epithelial dysplasia but which was later reported to be normal tissue and two patients had negative frozen sections that were later reported as showing squamous epithelial dysplasia on the paraffin sections. The Sensitivity was 92.8%, the Specificity 99.8% and the Kappa value was calculated to be 0.943 (Kappa values between 0.81–1.00 show almost perfect agreement).
Stage of tumour
Number of patients (n=82)
I II III IV
14 15 19 34
(17%) (18%) (23%) (42%)
Table 3. Site of tumour
Site of tumour Tongue Floor of mouth Lower alveolus and mandible Buccal mucosa Palate Retromolar area Lip Other
Number of patients (n=82) 25 23 8 6 5 5 3 7
Nine patients who had frozen sections reported as containing dysplastic squamous epithelium or invasive tumour underwent further local resection (Table 4); excision was then judged to be complete in eight of these patients with the help of further frozen sections. Tissue that was resected because of positive frozen sections was found to be positive for invasive tumour in three patients and for dysplasia in one patient. This tissue was found to be negative in four of these cases. In one patient (case 8, Table 4) it appears the tissue was inadvertently processed with the main specimen and this has therefore not been reported. In case 2 (Table 4) although two deep frozen sections were positive for tumour intraoperatively and tissue subsequently resected was positive for tumour, the deep final resection margins appeared free from tumour as far as could be assessed. There was extensive vascular permeation by the tumour in the specimen and lymph nodes in the specimen were positive with extracapsular spread. Similarly in case 5 (Table 4) further local resection was carried out because frozen section identified tumour confirmed to be an incidental neurofibroma later in paraffin preparation. Further resected tissue, however, was positive with a focus of carcinoma. Finally, one patient (case 3, Table 4) had an extensive T4 lesion in the maxilla with skull base involvement where tumour clearance was not possible. There were eight patients who had mild dysplasia on frozen sections intraoperatively but in whom further local resection was not carried out. Seven of
154 Ribeiro et al.
Table 4. Patients who had further local resection dictated by frozen sections
Patient
pTNM
Frozen section
Further local resection tissue
Main specimen margins Mucosal
Deep
Bone
Further management
1 2
T4N1M0 T4N1M0
2 Carcinoma in situ 2 Deep Frank tumour
Dysplasia Frank tumour
Dysplasia 4 mm
2 mm 7 mm
Clear —
Radiotherapy Radiotherapy
3 4
T4N2M0 T1N0M0
Frank tumour Negative
Frank tumour 10 mm
Frank tumour 10 mm
Frank tumour —
Radiotheraphy —
5
T2N2M0
7 mm
—
T4N0M0 T2N0M0
Ca in tissue, dysplasia at the margin Negative Negative
8 mm
6 7
1 Frank tumour 2 FS Dysplasia PS inflammation only FS reported positive found to be incidental neurofibroma on PS FS ?Reactive changes 3 Dysplasia
6 mm 5 mm
4 mm 7 mm
Clear —
Pt declined radiotheraphy Radiotherapy Radiotherapy
8 9
T4N2M0 T1N0M0
1 Frank tumour 1FS Dysplasia 2PS Dysplasia
— Negative
4 mm 7 mm
Frank tumour 8 mm
Clear —
Radiotherapy —
NSR—no sign of local recurrence, NCL—no palpable cervical lymphadenopathy, FS—frozen section, PS—paraffin section.
Local recurrence/ outcome NSR NCL NSR died of neck disease Yes died of disease NSR NCL NSR died of neck disease NSR NCL NSR died of neck disease NSR NCL NSR NCL Died of other disease
Frozen sections analysis
155
Table 5. Main specimen margins and outcome Frozen sections Margins
T Stage
Recurrence
No. of patients
Dysplastic
Ca in situ
Frank tumour
T1
T2
T3
T4
Local
Neck
Died of disease
50
6
—
—
11
22
5
12
2
14
13
7 8 5
1 — 2
— — 1
1 — —
1 — 1
— 3 2
1 2 1
5 3 1
— 2 2
2 2 1
1 2 1
3 7 1 1
— — 1 —
— — — —
— 1 — 1
1 — — —
— — — —
— 2 — —
2 6 1 1
3 3 1 1
1 2 1 1
2 3 1 1
Clear margins Close margins (<5 mm) Mucosal Deep Dysplasia/Ca in situ in mucosal margins Frank tumour at margin Mucosal Deep Bone Mucosal deep and bone
Fig. 1. Shows a close margin (<5 mm): invasive, rather poorly differentiated tumour extending to within about 1 mm (A) of the resection margin (B). H&E 50.
these patients had T1–T2 lesions where a wide local resection was performed and further local resection was considered impracticable. Five of these patients had clear final resection margins, two had dysplastic margins and one patient had bone involvement in the final surgical specimen. As shown in Table 5, in 15 (18%) patients the margins of the subsequent main resection specimen were judged to be close to the invasive tumour (<5 mm), Fig. 1, five (6%) patients had dysplasia and 12 (15%) patients had tumour at a resection margin. Of the 12 patients with frank tumour at a resection margin, seven had deep margins involved on the main specimen. Three of these patients had adjacent bone removed, which was clear and might
have represented the true resection margin. The patients were followed up for between 83 and 33 months postoperatively, mean: 55 months (SD 13.90), or until they died. There were 14 (17%) patients who developed local recurrence between 2 months to 46 months (median: 6.98 months) postoperatively. Of these, 11 (79%) had stage IV disease. Two (14%) patients had close (<5 mm) deep margins on the main surgical specimen and eight (57%) patients had invasive tumour at a resection margin. Of the eight patients with invasive tumour at the main specimen margin three had mucosal, three had deep tissue, one had bone invaded and the final patient, in whom clearance could not be achieved, had mucosal, deep and bone margins invaded by tumour. Four (29%) patients who developed local recurrence had resection margins that were negative. To date, 24 (29%) patients have died of their disease, of these five (21%) died as a result of recurrence at the primary site only, six (25%) died with both local and cervical/metastatic disease and 13 (54%) died of cervical/metastatic disease with no local recurrence. The median survival time for the patients who died was 11.78 months. In addition, 17 patients died from other causes.
Discussion The complete excision of tumour is an unquestionable principle of surgical oncology. The majority of patients that present to our unit, that have been represented in this study, have extensive disease at presentation (42% presented with stage IV disease in this study) and often have significant co-morbidity reflected by the proportion of patients
(21%) that died from other causes. W et al.21 have shown that survival is influenced by stage of disease, pattern of invasion and by the presence of tumour at or near the resection margin. Frozen sections may help improve the outcome of patients with margins involved by tumour. L11, in his study of 1522 cases of aerodigestive tract carcinoma, has shown that the oral cavity was the most likely site to have margins involved with residual tumour, a rate of 15% for margins positive for tumour, a figure similar to that found in this study. He recommended intraoperative control of margins with frozen sections and prompt re-excision for an optimal result, although only 50% of re-excisions were found to contain tumour. In our study nine (11%) patients had a further local resection based on the finding of invasive tumour or squamous epithelial dysplasia on frozen section analysis. Eight of these patients were subsequently judged to be clear of invasive tumour on further frozen sections taken at the time of surgery. The tissue subsequently resected was only evaluated for positivity—its margins could not be assessed because of problems with orientation and interpretation, particularly where deep tissue was resected. In four of eight cases i.e. 50% where this tissue was examined there was tumour involvement or dysplasia, a figure similar to that reported by L11 for re-excisions undertaken when resection margins were reported positive for tumour. Almost all these patients, Table 4, received postoperative radiotherapy and only the patient in whom clearance could not be achieved at the time of primary resection developed local recurrence. Three of these patients, however, died of neck disease. L et al.12 have shown a high rate of recurrence (74%) in close margins
156
Ribeiro et al.
Fig. 2. No epithelium identified on the frozen section. H&E 200.
Fig. 3. Paraffin preparation of the same sample shown in Fig. 2, which shows mild to moderate dysplasia. H&E 200.
<5 mm from tumour. Of the 82 patients in our study two (13%) of 15 patients with close margins developed local recurrence as opposed to 2 (4%) of 50 patients with clear margins. Frozen sections do not help improve close margins. Also allowances are not usually made for tissue shrinkage when the margins are judged by the surgeon. This may be interpreted as close margins, after tissue shrinkage, by the histopathologist. J et al.8 in their study on mongrel dogs have shown that the mean shrinkage from initial resection to final microscopic assessment of the lingual surface mucosal margins was 30.7% (3.7 mm), the deep tongue margin 34.5% (5.4 mm) and that for labiobuccal mucosal margins was 47.3% (5.7 mm). The greatest proportion of shrinkage occurred immediately upon resection, there being no disproportionate shrinkage observed. They recommend that an in situ margin of resection of at least 8 to 10 mm need to be taken to obtain 5 mm of pathologically clear margins.
Eight patients in our study had dysplastic frozen sections but further resection was not considered appropriate. Squamous epithelial dysplasia is often extensive with wide field changes often seen in patients with oropharyngeal carcinomas, making adequate excision impracticable in this group of patients. One of the purposes of this study was to assess whether there was failure to identify dysplastic epithelium or invasive carcinoma in frozen tissue. Histological examination of frozen tissue is now commonplace within surgical pathology practice in many units and plays an important role in decisions made by the surgeon during an operative procedure. The procedure is however time consuming, costly and sometimes stressful. It is also well known by surgeons and histopathologists that there are sampling errors and interpretational difficulties in the assessment of frozen tissue in which cellular morphology is not as clearly defined as in formalin fixed and then routinely processed tissue.
In this study resection margin tissue biopsies were examined particularly for the presence of epithelial dysplasia and invasive tumour. A potential difficulty in such biopsies is in differentiating between the reactive epithelial changes seen as a consequence of mucosal inflammation and true epithelial dysplasia. Another potential difficulty is that only a small number of sections of such biopsies can be taken due to the necessary constraints of time and it is therefore possible that dysplastic or neoplastic changes may not be sampled since they are present deeper within the frozen tissue block. This is a well recognized limitation of frozen section examination. Neither of these potential problems proved significant when the findings of frozen section were compared with those of subsequent paraffin embedded tissue. There was concordance in 548 of 551 biopsies, resulting in an overall accuracy of over 99.5%. Accuracy rates of between 96% and 99% have been previously reported6,7,11,17. Processing error occurred in one of the cases with no concordance between the cryostat and subsequent paraffin section (Figs 2 and 3) here the epithelium was not present in the cryostat section, missing dysplasia which was seen in the subsequent paraffin section. In the second case, although the frozen section was negative, a small area of dysplasia was detected in the subsequent paraffin section; a dysplastic region was therefore missed, as it was present in another region of the tissue. In the third case mucosal inflammation gave the impression of dysplasia in the frozen section, which was subsequently reported as normal tissue in the paraffin preparation (Figs 4 and 5). Despite the very satisfactory performance of frozen section reporting in this study, there has been failure to achieve adequate clearance of invasive tumour in 12 (15%) patients. Ten of these 12 patients who had invasive tumour at the final resection margin had clear frozen sections intraoperatively. Clearly the main issue is that the surgeon can take only a small number of biopsies from the resection margin, which represent only a very small percentage of the resection margin. Seven of these patients had invasive tumour at the deep margins. The deep margins are particularly difficult to assess intraoperatively and additional frozen sections taken from these margins might facilitate clearance in selected cases.
Frozen sections analysis
157
Frozen sections are accurate, however their shortfall is susceptibility to sampling errors. They are only as accurate as the specimen taken.
References
Fig. 4. Some epithelial cells have apparently enlarged nuclei, possibly also hyperkeratotic. This was therefore reported as showing mild to moderate dysplasia. H&E 200.
Fig. 5. Paraffin preparation of the same sample as shown in Fig. 4 shows epithelium that looks more orderly with chronic inflammatory cells producing reactive epithelial changes. There was no dysplasia. H&E 200.
Improved control has also been demonstrated following postoperative radiotherapy when there has been a positive surgical margin14,18,22. V et al.18 have shown that 10.5% of patients with positive margins treated with postoperative radiotherapy developed recurrence compared with 73% of patients who did not receive radiotherapy. Patients with tumour at a resection margin subsequently underwent radiotherapy; of these patients 8 developed local recurrence, and 7 patients died from their disease. Our findings concur with others with respect to subsequent local recurrence i.e. high incidence of recurrence in patients with positive margins on the final resection specimen (86%). However 2 of 50 patients (14%) with negative surgical margins on the main resection specimen developed local recurrence. Tumour recurrence can still occur
despite histologically clear margins. This may be because islands of tumour cells remain which cannot be detected microscopically. Brennon et al.1 have shown in a study using molecular techniques that clonal populations of infiltrating tumour cells harbouring mutations of the p53 were detected in 52% of specimens of their patients with histopathologically negative margins, and 38% of patients who had molecular evidence of such mutations at a surgical margin developed recurrence, compared with none in the negative margin group. Molecular biology, however, is not available perioperatively. The development of a more sensitive tool to ensure complete tumour excision is awaited. Frozen sections improve the prospect of the resection achieving tumour clearance and their routine use is justified. They do not influence the outcome of patients with close resection margins.
1. B AJ, M L, H RH, B JO, E YJ, K MW, G SN, S D. Molecular assessment of histopathological staging in squamous cell carcinoma of the head and neck. N Engl J Med 1995: 332: 429–435. 2. B RM, B KI, B B, L M. The prognostic and therapeutic value of frozen section determinations in the surgical treatment of squamous carcinoma of the head and neck. Am J Surg 1978: 136: 528. 3. C TY, E LJ, D DL. The clinical significance of pathological findings in surgically resected margins of the primary tumour in head and neck carcinoma. Int J Radiat Oncol Biol Phys 1987: 13: 833–837. 4. C JA, J AS, P DE, S L E. Implications of tumour in resection margins following surgical treatment of squamous cell carcinoma of the head and neck. Clin Otolaryngol 1993: 18: 37–41. 5. F MJ, L JD. The changing pattern of oral cancer 1977–1995. Br J Oral Maxillofac Surg 1995: 33: 328. 6. G-E R, D PJ, W D. The accuracy and clinical utility of frozen section diagnosis in head and neck surgery. Experience at a university medical centre. Head Neck 1993: 15: 33–38. 7. H PJ, H GG, Z RJ. The accuracy of frozen section diagnoses in 34 hospitals. Arch Pathol Lab Med 1990: 114: 355. 8. J RE, S JD, F GF, R RA, H HT. Quantification of surgical margin shrinkage in the oral cavity. Head Neck 1997: 19(4): 281– 286. 9. J AS, B H Z, N V, R NJ, K A, H TR. Do positive resection margins after ablative surgery for head and neck cancer adversely affect prognosis? A study of 352 patients with recurrent carcinoma following radiotherapy treated by salvage surgery. Br J Cancer 1996: 74: 128– 132. 10. K Z, L S, G B, D A. Frozen-section diagnosis in surgical pathology: a prospective analysis of 526 frozen sections. Cancer 1986: 57: 377. 11. L JG. Detection of residual carcinoma of the oral cavity, oropharynx, hypopharynx and larynx: a study of surgical margins. Trans Am Acad Ophthalmol Otolaryngol 1974: 78: 49–53.
158
Ribeiro et al.
12. L KG, S JP, S EW. The significance of ‘‘positive’’ margins in surgically resected epidermoid carcinomas. Head Neck Surg 1978: 1: 107–111. 13. L TR, S EW. Significance of positive margins in oral cavity squamous carcinoma. Am J Surg 1990: 160: 110–114. 14. M RVP, H RE, L EJ, et al. Postoperative radiotherapy for persistent tumour at the surgical margin in head and neck cancers. Laryngoscope 1983: 93: 1037–1040. 15. O RA, A S. Accuracy of frozen sections in assessing margins in oral cancer resection. J Oral Maxillofac Surg 1997: 55: 663–669. 16. R C, K EC, C P. Accuracy of frozen section diagnosis in a teaching hospital. Arch Pathol Lab Med 1987: 111: 564.
17. S P, B RM, B JG, W P, S H. Microscopic cutthrough of cancer in the surgical treatment of squamous carcinoma of the tongue. Am J Surg 1986: 152: 354–360. 18. V B, S EW, S JP, S R. Failure at the primary site following multimodality treatment in advanced head and neck cancer. Head Neck Surg 1984: 6: 720–723. 19. W JA. T2 carcinoma of the tongue: the histopathologists perspective. Br J Oral Maxillofac Surg 1999: 37: 187– 193. 20. W JA, R S, W CR, E ER, B JS, V ED. Survival and patterns of recurrence in 200 oral cancer patients treated by radical surgery and neck dissection. Oral Oncol 1999: 35: 257–265.
21. W JA, S J, V ED, B JS, W CR, R S. Survival, metastasis and recurrence of oral cancer in relation to pathological features. Ann R Coll Surg Engl 1995: 77: 325–331. 22. Z AL, J JT, M EN, T PB. Squamous cell carcinoma with positive margins. Arch Otolaryngol Head Neck Surg 1986: 112: 863–866. Address: Mr Noel F. F. Ribeiro Maxillofacial Unit North Manchester General Hospital Delaunays Road Crumpsall Manchester M8 5RB UK