Dermatologic I
surgery
II
II
I
I
III
The reliability of frozen sections in the evaluation of surgical margins for melanoma John A. Zitelli, M D , a Ronald L. Moy, MD, e and Edward Abell, M D b Pittsburgh,
Pennsylvania, and Los Angeles, California As the width of surgical margins declines, histologic evaluation of the margins is needed to assess the completeness of excision of a malignant melanoma. We studied 221 specimens in 59 patients and compared the interpretations of frozen and paraffin sections from the same block. Frozen sections had a sensitivity of 100% in detecting melanoma when present and a specificity of 90%. (J AM ACAD DV.RMATOL1991;24:102-6.) The surgical treatment of malignant melanoma ( M M ) is changing; narrow surgical margins are replacing wide excision because survival of patients with stage I M M is unrelated to the width of the surgical margin.l'25 As surgeons excise the narrowest margin needed to remove the tumor completely, they must rely more on histologic assessment of the margin for adequacy of excision. This study evaluates the reliability of frozen sections for evaluation of surgical margins in patients with M M . This information might be particularly useful during excision of M M with poorly defined margins or in areas where tissue conservation is particularly important. MATERIAL AND M E T H O D S Surgical margins were evaluated histologically in 59 consecutive patients with M M or lentigo maligna. Tumors were evaluated for thickness and depth of invasion. With the use of local anesthetic all melanomas were excised with a narrow margin of normal skin (6 ram) by the Mobs micrographic technique. The margins of the excised specimen were evaluated by tangential frozen sections. If tumor was present at the margin, an additional margin was excised from the area found to contain tumor. The horizontal frozen sections of Mobs surgery allow examination of 100% of the lateral and deep margins whereas traditional "bread-loafing" techniques of standard histopathology examine approximately 0.1% of the
margin. 26 In this study, horizontal Mohs surgery specimens were obtained by flattening the tissue on a glass slide while freezing in a cryostat. Care was taken to freeze the specimen quickly and to avoid refreezing to prevent freeze-induced artifact. Sections were cut at 4/zm and stained with hematoxylin and eosin. Specimens were evaluated immediately by the authors (J. A. Z., R. L. M.). Only the presence or absence of melanoma was reported. Melanoma was identified as nests of atypical cells within the dermis or epidermis or as mclanocytic hypcrplasia. Single isolated atypical melanocytes were not identified as melanoma at the margin because they are commonly found in normal skin adjacent to many benign and malignant skin disorders as well as in actinically damaged skin. 27 The distinction betweenatypical mclanocytichyl~rplasia scored as melanoma and scattered atypical mclanocytes scored as benign required the examination of three or more serial sections or deeper cuts. When the diagnosis was not immediately clear review of serial or deepcr sections usually showed enough variability to sway the decision. Often early multiple adjacent atypical cells or nesting was seen in true melanoma or further review showed more sparse atypia. If the diagnosis could not be reached after review of other sections, the equivocal findings were marked as positive to be sure that rccxcision completely removed the tumor. Usually no attempt was made to distinguish benign from malignant melanocytic lesions; thus even benign ncvi at the margins were excised. Areas of suspected regression of melanoma (dermal melanophages, fibrosis and a lymphohistocyfic infiltrate) were also marked positive although no melanoma was seen.
From the Departments of Medicine and Surgery, Montefiore Hospital, University Health Center of Pittsburgha; the Department of Dermatology, University of Pittsburghb; and the Department of Dermatology, University of California, Los AngelesY Reprint requests: John A, Zitelli, MD, Suite 10ll, 3471 Fifth Ave., Pittsburgh, PA 152/3.
102
After the results were recorded, the frozen tissue blocks were processed for paraffin sectioning. The flat surfaces of the tissue cut by frozen section were the same surfaces sectioned after paraffin embedding. These specimens were evaluated by a dermatopathologist (E. A.) in a blinded fashion. The results from frozen and paraffin sections were compared.
Volume 24 Number 1 January 1991
Frozen sectionsfor evaluation of surgical margins of melanoma 103
Table I. Specificity and sensitivity of frozen
Table II. Melanoma thickness
sections (ram)
Frozen sections
Paraffin sections
Positive
Negative
Positive Negative
18 19
0 179
I
[
Total
18 198
Sensitivity (the probabilitythat a patient with melanoma tests positive) (F+/P+) is 18/18 or 100%. Specificity (the probability that a patient free of melanoma tests negative ( F - / P - ) is 179/198 or 90%.
RESULTS
The quality of paraffin sections made from previously frozen tissue was excellent. Five specimens were eliminated from calculations of sensitivity and specificity because of inadequate paraffin sections. In no specimen was melanoma missed on frozen section when present on paraffin section. Frozen sections correlated with paraffm sections in 197 of the 221 specimens (48 patients) (Table I). This ineluded 179 specimens negative for melanoma on frozen and paraffin sections, and 18 specimens positive for melanoma on frozen and paraffin sections. The remaining 24 specimens ( 11 patients) were interpreted differently by frozen and paraffin sections. In 16 specimens (four patients) the frozen section interpretation was incorrect when compared with paraffin sections. These specimens were read as melanoma or regressing melanoma by frozen section but paraffin section showed only sun-induced epidermal atypia without melanoma. Three specimens (three patients) with different interpretations showed nests of melanocytes on frozen section and were interpreted as benign junctional nevi on paraffin section. However, we did not attempt to distinguish benign from malignant melanoeytic lesions by frozen section, although they were counted as incorrect readings for determining the sensitivity of the frozen-section technique. Five specimens (five patients) were eliminated from calculations of sensitivity and specificity. Four specimens (four patients) were interpreted as positive for melanoma by paraffin section but negative by frozen section. However, these specimens clearly represented false-positive paraffin readings because sections were made too deeply into the paraffin blocks. This was documented by the pronounced alteration in size and shape of the paraffin section compared with the frozen section, or by the appear-
In situ 3.50 Total
I
No. of
[
patients
19 18 8 10 4 59
ance of islands of epidermis that contained melanoma within the deep margin of the paraffin section in which only fat was seen in the frozen section. These were sections from difficult specimens in which melanoma had not been completely debulked surgically. These four patients with false-positive paraffin sections did not have reexeision after interpretation of paraffin sections. After follow-up of 36 to 43 months (average 39 months), there is no evidence of recurrence or metastases in these four patients; this further confirms that the frozen sections were not incorrect. One other specimen (one patient) eliminated from calculation of specificity and sensitivity had melanoma on frozen section but not on paraffin section. The frozen section was believed to be correct because reexcision of the positive area of the wound margin still showed melanoma by both paraffin and frozen section techniques. No patient with an incorrect interpretation by frozen or paraffin section had more than one additional 3 mm margin of skin excised from a portion of the wound, and in no patient did this additional margin alter the choice of wound management or affect the cosmetic or functional result. The surgical margin required to remove the melanoma completely was never greater than 1.5 cm. The thickness of the melanomas in these 59 patients is shown in Table II. Forty-nine patients had primary melanomas and 10 patients had recurrent melanomas. The duration of follow-up ranged from 31 t 62 months (average 41 months). There were no local recurrences in any of the 59 patients. DISCUSSION In the past when the standard surgical margin for melanoma was 5 cm, histologic evaluation of the margins was unnecessary because it was unlikely
104
ZitellietaL
Journal of the American Academy of Dermatology
Fig. 1. A recurrent lentigo maligna melanoma with poorly defined clinical margins. Standard surgical procedure would require very wide margins to ensure complete excision. Fig. 2. Surgical wound during excision (Mohs micrographic surgery) shows the map that demonstrates in red where melanoma is still present at margin as determined by frozen section. No tumor was visible by clinical examination in these areas. Fig. 3. Photomicrograph of frozen section from tissue specimen 2 in Fig. 2 shows nests of melanoma cells. (Hematoxylin-eosin stain; •
that tumor extended to the margins. Now that is is common practice to excise melanoma with margins of 2 cm or less, it is important to assess the adequacy of excision by examining the margins of the specimen. This is particularly important for melanomas with poorly defined margins or melanomas excised with narrow margins to conserve tissue for reconstruction (Figs. 1, 2, and 3). The use of frozen section to detect melanoma requires experience. In frozen sections, epidermal melanocytes lack the shrinkage artifact seen in formalin-fixed paraffin sections. Instead, melanoma cells may be plump with a prominent cytoplasm that resembles keratinocytes, or they may be small and round or spindle-shaped. 2s Architectural pattern is important to detect areas of melanocytic hyperpla-
sia or nesting. The only significant difficulty in frozen section interpretation was detecting in situ melanoma in severely sun-damaged skin and in some mucous membranes. In severely sun-damaged skin, atypical keratinocytes and distortion of normal epidermal architecture blurred the distinction between in situ melanoma and in situ squamous cell carcinoma. Occasionally, difficulties were noted in interpreting early melanocytic hyperplasia in frozen sections of conjunctiva and nasal mucosa, although even smaU nests of in situ melanoma were easily identified. Technical preparation is important for correct interpretation. Sections need to be cut thin (4 to 6 ~zm) and carefully processed to avoid freezing artifact. Thicker frozen sections or those with any freeze ar-
Volume 24 Number 1 January 1991
Frozen sections for evaluation of surgical margins of melanoma 105
tifact cannot be accurately interpreted. We routinely examined at least three serial sections of each specimen. No difficulties were noted in interpreting invasive melanoma, even in sun-damaged skin or mucosa. Desmoplastic melanoma and perineural invasion were accurately identified in frozen sections. In one case of a recurrent desmoplastic melanoma, the frozen sections were compared not only with paraffin sections but also with S-t 00 monoclonal antibodystained permanent sections, and 100% sensitivity was also achieved. Interpretation of margins by frozen section is far easier than diagnosis of an unknown disease by frozen section. 29 For diagnosis of melanoma, frozen section is reported to be 95% to 99% accurate by experienced pathologists. 3~ The major difficulties are not the inability to see melanocytic disorders but rather to distinguish benign from malignant lesions (melanoma vs nevi). However, once a diagnosis has been made, the use of frozen sections for margin evaluation becomes far simpler. This was emphasized in a study that compared 482 frozen sections with paraffin sections of various diseases. 29 Frozen section was 100% accurate in evaluating the resection margins and 97% accurate in identification of "correct pathologic process" (i.e., benign vs malignant). However, frozen sections were only 83% accurate for precise diagnosis. The authors emphasized that when the pathologist evaluates frozen sections for tumor margins, he knows the diagnosis and knows what to look for, a different and simpler task than evaluating frozen sections for a diagnosis of an unknown disease. Studies of melanomas excised by Mobs micrographic surgery document the accuracy of frozen sections during excision of melanoma with narrow surgical margins. The cure rate and recurrence rate with frozen section control is at least as good as surgery with wide margins. 36-41 The results of this study have direct clinical implications. The high sensitivity and specificity of frozen sections document its reliability during surgical excision. This might be particularly useful during excision of melanomas with poorly defined margins (lentigo maligna and lentigo maligna melanoma, amelanotic melanoma, desmoplastic melanoma, subungual melanoma, and mucosal melanoma) (Figs. 1, 2, and 3). Frozen section control would also help to guide the surgeon in cases in which conservation of tissue around a melanoma is
important for cosmetic and functional reasons (melanomas on the face, digits, or genitalia). The limitations of frozen sections diminish their use in some instances. Melanomas that arise in giant congenital nevi represent a case in which frozen sections might not be specific enough to distinguish benign melanocytes from malignant ones. Difficulties might also arise in melanomas that arise in skin with severe actinic keratoses, acute dermatitis, or large areas of hyperpigmented scar tissue. In these cases, questionable frozen sections should be confirmed by paraffin sections, although similar difficulties exist with interpretation of paraffin sections. Interpretation of frozen sections of melanoma surgical margins is accurate when sections are of good quality and interpretation is done by experienced reviewers. The high sensitivity (tack of falsenegatives) by frozen section emphasizes its accuracy not only during Mohs micrographic surgery but also during standard excision of difficult melanomas. It makes sense to examine surgical margins when excisions with narrow margins are attempted. Failure to examine the margin by intraoperative frozen sections may risk leaving melanoma behind or delaying the results of margin examination until parafl]n sections are done. REFERENCES
1. Olsen G. The malignant melanoma of the skin--new theories based on a study of 500 cases. Acta Chir Scand Suppl 1966;365:1-220. 2. Breslow A, Macht SD. Optimal size of resection margin for thin cutaneous melanoma. Surg Gynecol Obstet 1977;145:69l-2. 3. Bagley FH, Cady B, Lee A, et al. Changes in clinical presentation and management of malignant melanoma. Cancer 1981;47:2126-34. 4. Balch CM, Murad TM, Soong S J, et at. Tumor thickness as a guide to surgical management of clinical stage I melanoma patients. Cancer 1979;43:883-8. 5. Cascinelli N, Van der Esch EP, Breslow A, et al. Stage I melanoma of the skin: the problem of resection margins. Eur J Cancer 1980;16:1079-85. 6. Cosimi AB, Sober A J, Mihm MC, et al. Conservative surgical management of superficially invasive cutaneous melanoma. Cancer 1984;53:1256-9. 7. Day CL, Mihm ML, Sober A J, et at. Narrower margins for clinical stage I malignant melanoma. N Engl J Med 1982;306:479-81. 8. Day CL, Lew RA. Malignant melanoma prognostic factors 3: surgical margins. J Dermatol Surg Oncol 1983;9:797801. 9. Eider DE, Guerry D, Heiberger R.M, et al. Optimal resection margin for cutaneous malignant melanoma. Plast Reconstr Surg I983;71:66-72. 10. Elias EG, Didolkar MS. A elinicopathologic study of prognostic factors in cutaneous malignant melanoma. Surg Gynecol Obstet 1977;144:327-34.
106 Zitelli et al. 11. Fisher JC. Safe margins for melanoma excision. Ann Plast Surg 1985;14:158-61. 12. Goldman LI, Byrd R.. Narrowing resection margins for patients with low risk melanoma. Am J Surg 1988; 155:242-4. 13. Hansen MG, McCarter AB. Tumor thickness and lymphocytic infiltration in malignant melanoma of the head and neck. Am 3 Surg 1974;128:557-61. 14. Heenan PJ, Weeramanthri T, Holman CDJ, et al. Surgical treatment and survival from cutaneous malignant melanoma. Aust N Z J Surg 1985;55:229-34. 15. Lang NP, Stair JM, Degges RD, et al. Melanoma today does not require radical surgery. Am J Surg 1984; 148:7236. 16. Schmoeekel C, Bockelbrink A, Bockelbrink H, et al. Low and high risk malignant melanoma-III: prognostic significance of the resection margin. Eur J Cancer Clin Oncol 1983;19:245-9. 17. Taylor BA, Hughes LE. A policy of selective excision for primary cutaneous malignant melanoma. Eur J Surg Oncol 1985;11:7-13. 18. Veronesi V, Cascinelli N, Adamus J, et al. Thin stage 1 primary cutaneous malignant melanoma. N Engl J Med 1988;318:1159-62. 19. KeUey JW, Sagebiel RW, Calderon W, et al. The frequency of local recurrence and microsatellites as a guide to reexeision margins for cutaneous malignant melanoma. Ann Surg 1984;200:759-63. 20. Rampen F. Melanoma of the skin: the problem of resection margins. Eur J Cancer 1981;17:589-90. 21. Sondergaard K, Schou G. Survival with primary cutaneous malignant melanoma, evaluated from 2012 cases: a multivariate regression analysis. Virch Arch (Pathol Anat) 1985;406:179-95. 22. Urist MM, Baleh CM, Soong S J, et al. The influence of surgical margins and prognostic factors predicting the risk of local recurrence in 3445 patients with primary cutaneous melanoma. Cancer t985;55:1398-402. 23. Milton GW, Shaw HM, McCarthy WH. Resection margins for melanoma. Aust N Z J Surg 1985;55:225-8. 24. Aitken DR, Clausen K, Klein JP, et al. The extent of primary melanoma excision. Ann Surg 1983;198:634-41. 25. Zeitels J, La Rossa D, Hamilton R, et al. A comparison of local recurrence and resection margins for stage I primary cutaneous melanomas. Plast Reconstr Surg 1988;81:68893.
Journal of the American Academy of Dermatology
26. Abide JM, Nahai F, Bennett RG. The meaning of surgical margins. Plast Reconstr Surg 1984;73:492-6. 27. Montagna W, Kirchner S, Carlisle K. Histology of sundamaged human skin. J AM ACAD DERMATOL 1989; 21:907-18. 28. McGovern VJ. Primary treatment ofmalignant melanoma. In: Melanoma, histological diagnosis and prognosis. New York: Raven Press, 1983:186-91. 29. Sawady J, Berner J J, Siegler EE. Accuracy of and reasons for frozen sections: a correlative, retrospective study. Hum Pathol 1988;19:1019-23. 30. Davis N. Melanoma. Curr Probl Surg 1976;13:1-63. 31. Little JH, DavidNC. Frozenseetiondiagnosisofsuspected malignant melanomas of the skin. Cancer 1974;34:116372. 32. Milton GW, Jelihovsky T. Frozen section examination in the diagnosis of cutaneous malignant melanoma (melanoblastoma). Med J Austral 1962;49:503-4. 33. Shafir R, Hiss J, Tsur H, et al. Pitfalls in frozen section diagnosis of malignant melanoma. Cancer 1983;51:1168-70. 34. Hirst E, Cains GD, Bale PM, et al. Diagnosis by frozen section examination. II. Results in skin lesions. Aust NZ J Surg 1969;38:216-20. 35. Davis NC. Surgical treatment of benign nevi. Current Probl Surg 1976;5:25-7. 36. Mohs FE. Fixed tissue micrographic surgery for melanoma of the ear. Arch Otolaryngol Head Neck Surg 1988; 114:625-31. 37. Mohs FE. Microscopically controlled surgery for periorbital melanoma: fixed-tissue and fresh-tissue techniques. J Dermatol Surg Oncol 1985;11:284-91. 38. Mohs FE. Chemosurgery for melanoma. Arch Dermatol 1977;113:285-91. 39. Zitelli JA, Mobs FE, Larson P, et al. Mohs micrographic surgery for melanoma. Dermatol Clin 1989;7:833-43. 40. Fewkes J, Mohs FE. Microscopically controlled surgical excision (the Mohs technique). In: Fitzpatrick TB, Eisen AZ, Wolff K, et al, eds. Dermatology in general medicine. New York: McGraw-Hill, 1987:2557-63. 41. Mohs FE. Chemosurgery: microscopically controlled surgery for skin cancer. Springfield, Ill: Charles C Thomas, 1978:225-48.
surgery
II
II
I
I
III
The reliability of frozen sections in the evaluation of surgical margins for melanoma John A. Zitelli, M D , a Ronald L. Moy, MD, e and Edward Abell, M D b Pittsburgh,
Pennsylvania, and Los Angeles, California As the width of surgical margins declines, histologic evaluation of the margins is needed to assess the completeness of excision of a malignant melanoma. We studied 221 specimens in 59 patients and compared the interpretations of frozen and paraffin sections from the same block. Frozen sections had a sensitivity of 100% in detecting melanoma when present and a specificity of 90%. (J AM ACAD DV.RMATOL1991;24:102-6.) The surgical treatment of malignant melanoma ( M M ) is changing; narrow surgical margins are replacing wide excision because survival of patients with stage I M M is unrelated to the width of the surgical margin.l'25 As surgeons excise the narrowest margin needed to remove the tumor completely, they must rely more on histologic assessment of the margin for adequacy of excision. This study evaluates the reliability of frozen sections for evaluation of surgical margins in patients with M M . This information might be particularly useful during excision of M M with poorly defined margins or in areas where tissue conservation is particularly important. MATERIAL AND M E T H O D S Surgical margins were evaluated histologically in 59 consecutive patients with M M or lentigo maligna. Tumors were evaluated for thickness and depth of invasion. With the use of local anesthetic all melanomas were excised with a narrow margin of normal skin (6 ram) by the Mobs micrographic technique. The margins of the excised specimen were evaluated by tangential frozen sections. If tumor was present at the margin, an additional margin was excised from the area found to contain tumor. The horizontal frozen sections of Mobs surgery allow examination of 100% of the lateral and deep margins whereas traditional "bread-loafing" techniques of standard histopathology examine approximately 0.1% of the
margin. 26 In this study, horizontal Mohs surgery specimens were obtained by flattening the tissue on a glass slide while freezing in a cryostat. Care was taken to freeze the specimen quickly and to avoid refreezing to prevent freeze-induced artifact. Sections were cut at 4/zm and stained with hematoxylin and eosin. Specimens were evaluated immediately by the authors (J. A. Z., R. L. M.). Only the presence or absence of melanoma was reported. Melanoma was identified as nests of atypical cells within the dermis or epidermis or as mclanocytic hypcrplasia. Single isolated atypical melanocytes were not identified as melanoma at the margin because they are commonly found in normal skin adjacent to many benign and malignant skin disorders as well as in actinically damaged skin. 27 The distinction betweenatypical mclanocytichyl~rplasia scored as melanoma and scattered atypical mclanocytes scored as benign required the examination of three or more serial sections or deeper cuts. When the diagnosis was not immediately clear review of serial or deepcr sections usually showed enough variability to sway the decision. Often early multiple adjacent atypical cells or nesting was seen in true melanoma or further review showed more sparse atypia. If the diagnosis could not be reached after review of other sections, the equivocal findings were marked as positive to be sure that rccxcision completely removed the tumor. Usually no attempt was made to distinguish benign from malignant melanocytic lesions; thus even benign ncvi at the margins were excised. Areas of suspected regression of melanoma (dermal melanophages, fibrosis and a lymphohistocyfic infiltrate) were also marked positive although no melanoma was seen.
From the Departments of Medicine and Surgery, Montefiore Hospital, University Health Center of Pittsburgha; the Department of Dermatology, University of Pittsburghb; and the Department of Dermatology, University of California, Los AngelesY Reprint requests: John A, Zitelli, MD, Suite 10ll, 3471 Fifth Ave., Pittsburgh, PA 152/3.
102
After the results were recorded, the frozen tissue blocks were processed for paraffin sectioning. The flat surfaces of the tissue cut by frozen section were the same surfaces sectioned after paraffin embedding. These specimens were evaluated by a dermatopathologist (E. A.) in a blinded fashion. The results from frozen and paraffin sections were compared.
Volume 24 Number 1 January 1991
Frozen sectionsfor evaluation of surgical margins of melanoma 103
Table I. Specificity and sensitivity of frozen
Table II. Melanoma thickness
sections (ram)
Frozen sections
Paraffin sections
Positive
Negative
Positive Negative
18 19
0 179
I
[
Total
18 198
Sensitivity (the probabilitythat a patient with melanoma tests positive) (F+/P+) is 18/18 or 100%. Specificity (the probability that a patient free of melanoma tests negative ( F - / P - ) is 179/198 or 90%.
RESULTS
The quality of paraffin sections made from previously frozen tissue was excellent. Five specimens were eliminated from calculations of sensitivity and specificity because of inadequate paraffin sections. In no specimen was melanoma missed on frozen section when present on paraffin section. Frozen sections correlated with paraffm sections in 197 of the 221 specimens (48 patients) (Table I). This ineluded 179 specimens negative for melanoma on frozen and paraffin sections, and 18 specimens positive for melanoma on frozen and paraffin sections. The remaining 24 specimens ( 11 patients) were interpreted differently by frozen and paraffin sections. In 16 specimens (four patients) the frozen section interpretation was incorrect when compared with paraffin sections. These specimens were read as melanoma or regressing melanoma by frozen section but paraffin section showed only sun-induced epidermal atypia without melanoma. Three specimens (three patients) with different interpretations showed nests of melanocytes on frozen section and were interpreted as benign junctional nevi on paraffin section. However, we did not attempt to distinguish benign from malignant melanoeytic lesions by frozen section, although they were counted as incorrect readings for determining the sensitivity of the frozen-section technique. Five specimens (five patients) were eliminated from calculations of sensitivity and specificity. Four specimens (four patients) were interpreted as positive for melanoma by paraffin section but negative by frozen section. However, these specimens clearly represented false-positive paraffin readings because sections were made too deeply into the paraffin blocks. This was documented by the pronounced alteration in size and shape of the paraffin section compared with the frozen section, or by the appear-
In situ 3.50 Total
I
No. of
[
patients
19 18 8 10 4 59
ance of islands of epidermis that contained melanoma within the deep margin of the paraffin section in which only fat was seen in the frozen section. These were sections from difficult specimens in which melanoma had not been completely debulked surgically. These four patients with false-positive paraffin sections did not have reexeision after interpretation of paraffin sections. After follow-up of 36 to 43 months (average 39 months), there is no evidence of recurrence or metastases in these four patients; this further confirms that the frozen sections were not incorrect. One other specimen (one patient) eliminated from calculation of specificity and sensitivity had melanoma on frozen section but not on paraffin section. The frozen section was believed to be correct because reexcision of the positive area of the wound margin still showed melanoma by both paraffin and frozen section techniques. No patient with an incorrect interpretation by frozen or paraffin section had more than one additional 3 mm margin of skin excised from a portion of the wound, and in no patient did this additional margin alter the choice of wound management or affect the cosmetic or functional result. The surgical margin required to remove the melanoma completely was never greater than 1.5 cm. The thickness of the melanomas in these 59 patients is shown in Table II. Forty-nine patients had primary melanomas and 10 patients had recurrent melanomas. The duration of follow-up ranged from 31 t 62 months (average 41 months). There were no local recurrences in any of the 59 patients. DISCUSSION In the past when the standard surgical margin for melanoma was 5 cm, histologic evaluation of the margins was unnecessary because it was unlikely
104
ZitellietaL
Journal of the American Academy of Dermatology
Fig. 1. A recurrent lentigo maligna melanoma with poorly defined clinical margins. Standard surgical procedure would require very wide margins to ensure complete excision. Fig. 2. Surgical wound during excision (Mohs micrographic surgery) shows the map that demonstrates in red where melanoma is still present at margin as determined by frozen section. No tumor was visible by clinical examination in these areas. Fig. 3. Photomicrograph of frozen section from tissue specimen 2 in Fig. 2 shows nests of melanoma cells. (Hematoxylin-eosin stain; •
that tumor extended to the margins. Now that is is common practice to excise melanoma with margins of 2 cm or less, it is important to assess the adequacy of excision by examining the margins of the specimen. This is particularly important for melanomas with poorly defined margins or melanomas excised with narrow margins to conserve tissue for reconstruction (Figs. 1, 2, and 3). The use of frozen section to detect melanoma requires experience. In frozen sections, epidermal melanocytes lack the shrinkage artifact seen in formalin-fixed paraffin sections. Instead, melanoma cells may be plump with a prominent cytoplasm that resembles keratinocytes, or they may be small and round or spindle-shaped. 2s Architectural pattern is important to detect areas of melanocytic hyperpla-
sia or nesting. The only significant difficulty in frozen section interpretation was detecting in situ melanoma in severely sun-damaged skin and in some mucous membranes. In severely sun-damaged skin, atypical keratinocytes and distortion of normal epidermal architecture blurred the distinction between in situ melanoma and in situ squamous cell carcinoma. Occasionally, difficulties were noted in interpreting early melanocytic hyperplasia in frozen sections of conjunctiva and nasal mucosa, although even smaU nests of in situ melanoma were easily identified. Technical preparation is important for correct interpretation. Sections need to be cut thin (4 to 6 ~zm) and carefully processed to avoid freezing artifact. Thicker frozen sections or those with any freeze ar-
Volume 24 Number 1 January 1991
Frozen sections for evaluation of surgical margins of melanoma 105
tifact cannot be accurately interpreted. We routinely examined at least three serial sections of each specimen. No difficulties were noted in interpreting invasive melanoma, even in sun-damaged skin or mucosa. Desmoplastic melanoma and perineural invasion were accurately identified in frozen sections. In one case of a recurrent desmoplastic melanoma, the frozen sections were compared not only with paraffin sections but also with S-t 00 monoclonal antibodystained permanent sections, and 100% sensitivity was also achieved. Interpretation of margins by frozen section is far easier than diagnosis of an unknown disease by frozen section. 29 For diagnosis of melanoma, frozen section is reported to be 95% to 99% accurate by experienced pathologists. 3~ The major difficulties are not the inability to see melanocytic disorders but rather to distinguish benign from malignant lesions (melanoma vs nevi). However, once a diagnosis has been made, the use of frozen sections for margin evaluation becomes far simpler. This was emphasized in a study that compared 482 frozen sections with paraffin sections of various diseases. 29 Frozen section was 100% accurate in evaluating the resection margins and 97% accurate in identification of "correct pathologic process" (i.e., benign vs malignant). However, frozen sections were only 83% accurate for precise diagnosis. The authors emphasized that when the pathologist evaluates frozen sections for tumor margins, he knows the diagnosis and knows what to look for, a different and simpler task than evaluating frozen sections for a diagnosis of an unknown disease. Studies of melanomas excised by Mobs micrographic surgery document the accuracy of frozen sections during excision of melanoma with narrow surgical margins. The cure rate and recurrence rate with frozen section control is at least as good as surgery with wide margins. 36-41 The results of this study have direct clinical implications. The high sensitivity and specificity of frozen sections document its reliability during surgical excision. This might be particularly useful during excision of melanomas with poorly defined margins (lentigo maligna and lentigo maligna melanoma, amelanotic melanoma, desmoplastic melanoma, subungual melanoma, and mucosal melanoma) (Figs. 1, 2, and 3). Frozen section control would also help to guide the surgeon in cases in which conservation of tissue around a melanoma is
important for cosmetic and functional reasons (melanomas on the face, digits, or genitalia). The limitations of frozen sections diminish their use in some instances. Melanomas that arise in giant congenital nevi represent a case in which frozen sections might not be specific enough to distinguish benign melanocytes from malignant ones. Difficulties might also arise in melanomas that arise in skin with severe actinic keratoses, acute dermatitis, or large areas of hyperpigmented scar tissue. In these cases, questionable frozen sections should be confirmed by paraffin sections, although similar difficulties exist with interpretation of paraffin sections. Interpretation of frozen sections of melanoma surgical margins is accurate when sections are of good quality and interpretation is done by experienced reviewers. The high sensitivity (tack of falsenegatives) by frozen section emphasizes its accuracy not only during Mohs micrographic surgery but also during standard excision of difficult melanomas. It makes sense to examine surgical margins when excisions with narrow margins are attempted. Failure to examine the margin by intraoperative frozen sections may risk leaving melanoma behind or delaying the results of margin examination until parafl]n sections are done. REFERENCES
1. Olsen G. The malignant melanoma of the skin--new theories based on a study of 500 cases. Acta Chir Scand Suppl 1966;365:1-220. 2. Breslow A, Macht SD. Optimal size of resection margin for thin cutaneous melanoma. Surg Gynecol Obstet 1977;145:69l-2. 3. Bagley FH, Cady B, Lee A, et al. Changes in clinical presentation and management of malignant melanoma. Cancer 1981;47:2126-34. 4. Balch CM, Murad TM, Soong S J, et at. Tumor thickness as a guide to surgical management of clinical stage I melanoma patients. Cancer 1979;43:883-8. 5. Cascinelli N, Van der Esch EP, Breslow A, et al. Stage I melanoma of the skin: the problem of resection margins. Eur J Cancer 1980;16:1079-85. 6. Cosimi AB, Sober A J, Mihm MC, et al. Conservative surgical management of superficially invasive cutaneous melanoma. Cancer 1984;53:1256-9. 7. Day CL, Mihm ML, Sober A J, et at. Narrower margins for clinical stage I malignant melanoma. N Engl J Med 1982;306:479-81. 8. Day CL, Lew RA. Malignant melanoma prognostic factors 3: surgical margins. J Dermatol Surg Oncol 1983;9:797801. 9. Eider DE, Guerry D, Heiberger R.M, et al. Optimal resection margin for cutaneous malignant melanoma. Plast Reconstr Surg I983;71:66-72. 10. Elias EG, Didolkar MS. A elinicopathologic study of prognostic factors in cutaneous malignant melanoma. Surg Gynecol Obstet 1977;144:327-34.
106 Zitelli et al. 11. Fisher JC. Safe margins for melanoma excision. Ann Plast Surg 1985;14:158-61. 12. Goldman LI, Byrd R.. Narrowing resection margins for patients with low risk melanoma. Am J Surg 1988; 155:242-4. 13. Hansen MG, McCarter AB. Tumor thickness and lymphocytic infiltration in malignant melanoma of the head and neck. Am 3 Surg 1974;128:557-61. 14. Heenan PJ, Weeramanthri T, Holman CDJ, et al. Surgical treatment and survival from cutaneous malignant melanoma. Aust N Z J Surg 1985;55:229-34. 15. Lang NP, Stair JM, Degges RD, et al. Melanoma today does not require radical surgery. Am J Surg 1984; 148:7236. 16. Schmoeekel C, Bockelbrink A, Bockelbrink H, et al. Low and high risk malignant melanoma-III: prognostic significance of the resection margin. Eur J Cancer Clin Oncol 1983;19:245-9. 17. Taylor BA, Hughes LE. A policy of selective excision for primary cutaneous malignant melanoma. Eur J Surg Oncol 1985;11:7-13. 18. Veronesi V, Cascinelli N, Adamus J, et al. Thin stage 1 primary cutaneous malignant melanoma. N Engl J Med 1988;318:1159-62. 19. KeUey JW, Sagebiel RW, Calderon W, et al. The frequency of local recurrence and microsatellites as a guide to reexeision margins for cutaneous malignant melanoma. Ann Surg 1984;200:759-63. 20. Rampen F. Melanoma of the skin: the problem of resection margins. Eur J Cancer 1981;17:589-90. 21. Sondergaard K, Schou G. Survival with primary cutaneous malignant melanoma, evaluated from 2012 cases: a multivariate regression analysis. Virch Arch (Pathol Anat) 1985;406:179-95. 22. Urist MM, Baleh CM, Soong S J, et al. The influence of surgical margins and prognostic factors predicting the risk of local recurrence in 3445 patients with primary cutaneous melanoma. Cancer t985;55:1398-402. 23. Milton GW, Shaw HM, McCarthy WH. Resection margins for melanoma. Aust N Z J Surg 1985;55:225-8. 24. Aitken DR, Clausen K, Klein JP, et al. The extent of primary melanoma excision. Ann Surg 1983;198:634-41. 25. Zeitels J, La Rossa D, Hamilton R, et al. A comparison of local recurrence and resection margins for stage I primary cutaneous melanomas. Plast Reconstr Surg 1988;81:68893.
Journal of the American Academy of Dermatology
26. Abide JM, Nahai F, Bennett RG. The meaning of surgical margins. Plast Reconstr Surg 1984;73:492-6. 27. Montagna W, Kirchner S, Carlisle K. Histology of sundamaged human skin. J AM ACAD DERMATOL 1989; 21:907-18. 28. McGovern VJ. Primary treatment ofmalignant melanoma. In: Melanoma, histological diagnosis and prognosis. New York: Raven Press, 1983:186-91. 29. Sawady J, Berner J J, Siegler EE. Accuracy of and reasons for frozen sections: a correlative, retrospective study. Hum Pathol 1988;19:1019-23. 30. Davis N. Melanoma. Curr Probl Surg 1976;13:1-63. 31. Little JH, DavidNC. Frozenseetiondiagnosisofsuspected malignant melanomas of the skin. Cancer 1974;34:116372. 32. Milton GW, Jelihovsky T. Frozen section examination in the diagnosis of cutaneous malignant melanoma (melanoblastoma). Med J Austral 1962;49:503-4. 33. Shafir R, Hiss J, Tsur H, et al. Pitfalls in frozen section diagnosis of malignant melanoma. Cancer 1983;51:1168-70. 34. Hirst E, Cains GD, Bale PM, et al. Diagnosis by frozen section examination. II. Results in skin lesions. Aust NZ J Surg 1969;38:216-20. 35. Davis NC. Surgical treatment of benign nevi. Current Probl Surg 1976;5:25-7. 36. Mohs FE. Fixed tissue micrographic surgery for melanoma of the ear. Arch Otolaryngol Head Neck Surg 1988; 114:625-31. 37. Mohs FE. Microscopically controlled surgery for periorbital melanoma: fixed-tissue and fresh-tissue techniques. J Dermatol Surg Oncol 1985;11:284-91. 38. Mohs FE. Chemosurgery for melanoma. Arch Dermatol 1977;113:285-91. 39. Zitelli JA, Mobs FE, Larson P, et al. Mohs micrographic surgery for melanoma. Dermatol Clin 1989;7:833-43. 40. Fewkes J, Mohs FE. Microscopically controlled surgical excision (the Mohs technique). In: Fitzpatrick TB, Eisen AZ, Wolff K, et al, eds. Dermatology in general medicine. New York: McGraw-Hill, 1987:2557-63. 41. Mohs FE. Chemosurgery: microscopically controlled surgery for skin cancer. Springfield, Ill: Charles C Thomas, 1978:225-48.
