Histologic pattern analysis of basal cell carcinoma

Journal of the American Academy of Dermatology

Schreiber et al. REFERENCES 1. American Cancer Society. Facts and figures. Atlanta: 1988. 2. Scotto J, Fears TR, Fraumeni JF Jr. Incidence ofnonmelanoma skin cancer in the United States. Bethesda, Md: National Institutes of Health, 1983; Publication No. 832433. 3. Moon TE, Davidson J, Schreiber MM, et al. Sunlight intensity and skin cancer incidence in Arizona. Abstract presented at Proceedings of the American Academy of Dermatology 1989. 4. Goldberg LH, Carter-Campbell S, Yiannias JA. The risk

5. 6. 7. 8.

of acquiring a new skin cancer in patients with a previously diagnosed skin cancer. Skin Cancer Foundation Journal, World Congress IlIon Cancers of the Skin, 1988:25-6. Robinson JK. Risk of developing another basal cell carcinoma. Cancer 1987;60:118-20. Epstein E. Value of follow-up after treatment of basal cell carcinoma. Arch Dermatol 1973;108:798-800. Kalbfleisch JD, Prentice RL. The statistical analysis of failure time data. New York: John Wiley & Sons, 1980. Population Statistics Unit. A demographic guide to Arizona, 1985. Phoenix: Arizona Department of Economic Security, 1985; report 14.

Histologic pattern analysis of basal cell carcinoma Study of a series of 1039 consecutive neoplasms Mack Sexton, MD,a,c Daniel B. Jones, MD,b and Mary E. Maloney, MDC

Hershey, Pennsylvania This study attempts to define histologic patterns in 1039 consecutive cases of basal cell carcinoma and to correlate these patterns with adequacy of margins of surgical excision. Five major histologic patterns were identified: nodular, 218 cases (21%); superficial, 181 cases (17%); micronodular, 151 cases (15%); infiltrative, 77 cases (7%); and morpheic, 11 cases (1 %). A mixed pattern (two or more major histologic patterns) was present in 401 cases (38.5%). Our study indicates that nodular and superficial basal cell carcinomas can be completely removed by simple surgical excision in a high percentage of cases (93.6% and 96.4%, respectively) whereas the micronodular, infiltrative, and morpheic basal cell carcinomas have a higher incidence of positive tumor margins (18.6%,26.5%, and 33.3%, respectively) after excision. Mixed patterns that consisted of combinations of the nodular, micronodular, or infiltrative types exhibited a behavior similar to the pattern that resulted in a greater chance of incomplete surgical removal. (J AM ACAD DERMATOL 1990;23:1118-26.) Basal cell carcinomas are by far the most common primary cutaneous malignant neoplasm. Most are relatively innocuous and result in infrequent patient morbidity or mortality. However, some may cause extensive tissue destruction and directly contribute to death through either local infiltration of vital structures or metastatic disease. Historically, basal cell carcinomas have been histologically segregated into undifferentiated and differentiated types. 1,2 Differentiated basal cell carciFrom the Department of Pathology, a The College of Medicine,b and the Department of Medicine; Division of Dermatology, M. S. Hershey Medical Center, The Pennsylvania State University. Accepted for publication Aprill, 1990. Reprint requests: Mack Sexton, MD, University of Florida Medical Center, P.O. Box 1-275, JHMHC, Gainesville, FL 32610-0275.

16/1/21315

1118

nomas are named on the basis of their resemblance to cutaneous appendages. Undifferentiated basal cell carcinomas lack these differentiating features. However, this histologic classification is inadequate because of difficulties in its uniform application to therapy and its acceptance by dermatopathologists. This study attempts to define the histologic patterns encountered in 1039 consecutive cases of basal cell carcinoma. These patterns were correlated with adequacy of surgical excision because adequacy of excision is an effective predictor of the tendency of basal cell carcinomas to recur. 3

MATERIAL AND METHODS The cases in this study were retrieved from the pathology and dermatopathology archives at the M. S. Hershey Medical Center. A total of 1039 consecutive primary

Volume 23 Number 6, Part I December 1990

Histologic pattern analysis of basal cell carcinoma 1119

Fig. 1. Nodular basal cell carcinoma demonstrates large, well-defined, rounded tumor nodules with peripheral palisading and myxoid stroma surrounding the neoplasm. This type is significantly associated with central necrosis. (X20.)

basal cell carcinomas diagnosed from Jan. 1, 1987 through June 1988 were studied. Material for this study included tissue from an outpatient dermatology referral service, an academic dermatology practice, and other surgical specialties such as general surgery, otorhinolaryngology, and plastic surgery. There were 467 excision specimens, 441 shave biopsy specimens, 130 punch biopsy specimens, and 1 curettage specimen examined. Cases in which a biopsy specimen preceded excision were discarded and only the histologic pattern of the excisional specimen was tabulated according to the design ofthis study, which prevented double counting. Therefore only one pathology specimen of each neoplasm was studied. Histologic patterns defined as follows were assigned on the basis of a consensus among the three authors. Nodular type. The nodular basal cell carcinoma is defined as a rounded mass of neoplastic cells with well-defined peripheral contours (Fig. 1). Peripheral palisading is well developed. Mucinous retraction clefts are usually prominent. The stroma tends to be slightly myxoid around the neoplasm. Superficial type. We defined the superficial type as one or more tumor foci that extended from the epidermis and/or adnexae and impinged on or entered the papillary dermis (Fig. 2). Tumor islands are relatively small and sometimes resemble the primordial epithelial germ of embryonic development or rudimentary hair bulb structures. Peripheral palisading is usually well defined and the peripheral contours are rounded and smooth. Retraction spaces may be prominent and the surrounding fibrous tissue is slightly myxoid and compacted. Rare small foci of neoplasm may appear detached from the epithelium and present in the underlying papillary dermis. For the purposes of this study only neoplasms that showed these his-

tologic features were classified as superficial. If the neoplasm failed to meet these criteria or prominent foci of tumor were no longer in continuity with the epithelium, then the neoplasm was classified as one of the other patterns. In addition, many neoplasms exhibited another predominant pattern centrally with features of a superficial pattern laterally. These neoplasms were placed into one of the other categories and were not reported as mixed-superficial type neoplasms. MicronoduJar type. We defined this pattern as small rounded nodules of neoplasm that represent a miniaturization of the nodular type; smaller tumor islands tend to approximate the size of hair bulbs (Fig. 3). Tumor contours are rounded, well demarcated, and usually demonstrate peripheral palisading. Mucinous retraction spaces may be present but are generally less conspicuous than in nodular neoplasms. The surrounding tissue may demonstrate a more collagenous appearance than the myxoid stroma characteristic of the nodular pattern. Infiltrative type. This pattern demonstrates tumor islands of varying size; some tumor islands are large and others are small and resemble morpheic (ype neoplasms. The major histologic feature of the infiltrative pattern (Fig. 4) is the irregular, jagged, and spiking configuration of the peripheral contours of tumor islands. Peripheral palisading is poorly developed and mucinous retraction spaces are uncommon. Tumor stroma is more likely to be slightly collagenous rather than myxoid. Morpheic type. In this type of neoplasm tumor islands are small, elongated, and have sharp, angulated ends. Neoplastic islands infiltrate as narrow strands or cords of cells (Fig. 5). Mucinous retraction spaces are nire. The surrounding collagen may appear sclerotic but the presence of this type of stroma is not an absolute criterion for

Journal of the American Academy of Dermatology

1120 Sexton et al.

Fig. 2. A and B. Superficial basal cell carcinoma is strictly defined as nodules or buds of neoplasm extending from the epithelium into the papillary dermis. Peripheral contours resemble those of the nodular and/or micronodular types. (A, X20; B, XlOO.)

diagnosis. Like superficial basal cell carcinoma, we derme this pattern strictly and diagnose it only when it exists in a pure form as described earlier. The small tumor islands that may compose a part of the morphologic spectrum of some infiltrative basal cell carcinomas can appear identical. Mixed type. By far the most common type of mixed pattern was nodular-micronodular (Fig. 6). This mixture of patterns may be explained by a lack of regulation of tumor growth centers that results in formation ofvariably sized, rounded nodules. The ends ofthis spectrum (i.e., the nodular vs the micronodular type) are easily appreciated. However, there is a significant percentage of basal cell carcinomas that demonstrate an imperceptible blending between nodular and micronodular types. Often the micronodular pattern tends to be prominently situated around the periphery of a nodular type neoplasm that resembles satellites oriented around a large central nodule. Statistical methods. The relation between histologic pattern and adequacy of excision was examined with

the chi-square test for association. Likewise, the secondary changes we studied were examined by the same statistical method for a positive, negative, or no association.

RESULTS

Six hundr~ thirty-eight neoplasms demonstrated a predominant histologic pattern and 401 neoplasms demonstrated a mixed pattern (Table I). Table II compares and contrasts adequacy of excision versus histologic pattern. Table III summarizes histologic pattern versus ulceration, pigmentation, and cystic change. Nodular type. Of the basal cell carcinomas 218 (21%) exhibited a predominantly nodular pattern. Of the 94 excisional specimens of nodular basal cell carcinoma, only six (6.4%) demonstrated surgical margins involved by neoplasm. Superficial type. The superficial pattern was ob-

Volume 23 Number 6, Part 1 December 1990

Histologic pattern analysis of basal cell carcinoma 1121

Fig. 3. A and B. Micronodular basal cell carcinoma consists of small nodules of tumor demonstrating a tendency to approximate the size of hair bulbs. Peripheral palisading may be less well developed and the stroma may appear slightly more fibrous around the neoplasm. (A, X20; B, x400.)

served in 181 specimens (17.4%). Of the 56 excisional biopsy specimens of superficial basal cell carcinoma, only two (3.6%) demonstrated a positive margin. Superficial basal cell carcinomas were the ones most frequently pigmented (7.2%) and 17.1% are ulcerated. Cystic changes were infrequent (0.55%), probably because tumor islands were mostly small with little tendency for central avascular tumor necrosis. Micronodular type. One hundred fifty-one neoplasms in this series (14.5%) demonstrated a predominant micronodular pattern. Of 59 excisional specimens of the micronodular pattern, 11 (18.6%)

demonstrated positive margins. Six neoplasms were pigmented (4.0%), eight demonstrated cystic change (5.3%), and 44 were ulcerated (29.1%). In addition, we found that the micronodular basal cell carcinoma (or the formation of a micronodule) is most likely to undergo dystrophic calcification in basal cell carcinoma. Infiltrative type. The infiltrative basal cell carcinoma was identified in 77 of the 1039 cases examined (7.4%). Of 34 excisional specimens of infiltrative basal cell carcinoma, nine (26.5%) demonstrated tumor involvement of the surgical margins. Ulceration in the infiltrative basal cell car-

1122 Sexton et al.

Journal of the American Academy of Dermatology

Fig. 4. A, B, and C. Infiltrative basal cell carcinoma shows irregular, spiking tumor islands of varying size. Contours are irregular with little to no peripheral palisading of nuclei. This type rarely shows central necrosis of tumor islands or pigmentation in its pure form. (A, X40; B, X200; C, X400.)

cinoma was fairly cornman (33.7%). None of the lesions that demonstrated an infiltrative pattern was pigmented and none demonstrated cystic change. Morpheic type. The morpheic (or morpheaform) pattern was identified in .11 cases (1.1 %). Of the six excisional specimens of morpheic growth type basal cell carcinoma, two (33.3%) demonstrated positive margins. In addition, none oftheseneoplasms showed cystic change or pigmentation. Ulceration was presentiri two (18.2%). Mixed type.. A mixed pattern was prominent in 401 neoplasms in this series (38.5%). Of 115 excisions of the nodular-micronodular pattern, 19 (16.5%) demonstrated positive surgical margins. This finding closely approximates the positive margins in the micronodular type (18.6%). Seventy-nine neoplasms were Ulcerated, 28 were pigmented, and 71 showed cystic change. The other mixed patterns consisted of variations of nodular or micronodular

patterns with the infiltrative pattern. One hundred three excisions demonstrated this mixture of patterns. Positive margins were present in 31 (30.1 %). Of this category of neoplasms, 82 were ulcerated, 6 were pigmented, and 28 had cystic change. Statistical results. Overall there is a significant association between histologic pattern and adequacy of excision (p < 0.001). Specifically, nodular and superficial patterns as a group have a significantly lower percentage of positive margins compared with micronodular, infiltrative, and mixed with infiltrative patterns as a group (p < 0.001). The number of excisions of the morpheic type of basal cell carcinoma was too small to allow for this comparison to be made. Nodular basal cell carcinoma is positively associated with ulceration (p = 0.008). According to the estimated odds ratio, this neoplasm is 1.6 times more likely to be ulcerated when compared with other patterns. In contrast, the micronodular pattern

Volume 23 Number 6, Part I December 1990

Histologic pattern analysis of basal cell carcinoma 1123

Fig." 5. A and B. Morpheic basal cell carcinoma demonstrates prominent linear cords and strands of neoplasm. Mucinous retraction spaces are uncommon. These neoplasms also rarely demonstrate necrosis of tumor islands or pigmentation. (A, X20; B, x400.)

demonstrates a negative association with ulceration (p < 0.001). The estimated odds ratio for this neoplasm indicates that it is only 0.4 times as likely to be associated with ulceration when compared with other patterns. Superficial, infiltrative, and morpheic patterns were not associated with ulceration (p = 0.70, p = 0.22, and p = 0.47, respectively)'. There was no positive association of pigmentation with pattern. The infiltrative pattern demonstrates a negative association with pigmentation (p = 0.015). The other patterns demonstrated no association with pigmentation. With regard to cystic change, only the nodular basal cell carcinoma showed a positive association with this finding (p < 0.001). The estimated odds ratio indicates that this neoplasm is 35.27 times more likely to demonstrate cystic change

when compared with other types. Micronodular, superficial, and infiltrative type basal cell carcinomas demonstrated a negative association with cystic change (p < 0.001 for each). DISCUSSION

Thackray 4 was the first to correlate the histologic classification of basal cell carCinomas with prognosis. He categorized tumors as circumscribed, predominantly circumscribed with early infiltration, and infiltrative types and concluded that the extent of infiltration affected prognosis. Sloane5 also emphasized recognition of growth patterns in basal cell carcinoma in predicting recurrence. In his study of 156 basal cell carcinomas the nodular pattern showed only a 6% recurrence rate in contrast to the

Journal of the American Academy of Dermatology

1124 Sexton et at.

Fig. 6. Nodular-micronodular basal cell carcinoma is the most common mixed-pattern

basal cell carcinoma, which demonstrates features of each type. Other mixed-type basal cell carcinomas exist. (X20.)

Table I. Predominant and mixed histologic patterns of basal cell carcinomas Predominant pattern Nodular Superficial Micronodular Infiltrative Morpheic Subtotal Mixed pattern Nodular-micronodular Nodular-infiltrative Nodular-micronodular-infiltrative Micronodular-infiltrative Subtotal Total

No.

%

218 181 151 77 11 638

21.0 17.4 14.5 7.4 1.1 61.41

226

21.8

41 401 1039

3.9 38.6

73 61

7.0 5.9

25% rate of recurrence in infiltrative neoplasms. J acobs et al. 6 also considered the nodular basal cell carcinoma to be less invasive and less likely to recur in contrast to the ulcera6ve and infiltrative types. We also conclude that the nodular basal cell carcinoma is less likely to invade deeply and recur than other histologic types. The secondary changes in nodular basal cell carcinoma consisted of cyst formation, ulceration, and pigmentation. We conclude that these changes are unimportant in determining the biologic behavior of basal cell carcinoma and these terms should be

eliminated. Nodulocystic, noduloulcerative, and nodulopigmented are adjectives that have no prognostic value and these secondary changes may also occur in superficial and micronodular basal cell carcinomas. The micronodular basal cell carcinoma, when it composes all of the tumor bulk or is present as a prominent peripheral extension from the main body of an otherwise nodular type basal cell carcinoma, is more likely to be incompletely removed. These neoplasms have a less well-defined clinical border, which probably explains their higher rate of recurrence after excision. Lang and Maize? emphasized the importance of recognizing the micronodular basal cell carcinoma as a pattern more prone to recurrence. The recently described basal cell carcinoma with follicular differentiation displays a prominent micronodular pattern8 and we believe that the micronodular basal cell carcinoma is the type most likely to be confused with benign follicular neoplasms. Wermuth and Fajard09 reviewed the literature and concluded that no particular histologic pattern was characteristic ofmetastasizing basal cell carcinoma. However, the published photograph of their case shows a microriodular type neoplasm. The superficial basal cell carcinoma was the second most common type in this study. We define this pattern strictly and use this term only when the neoplasm exists in a pure histologic pattern. We no longer use the term multicentric because the neoplasm is probably not multicentric in its his-

Volume 23 Number 6, Part 1 December 1990

Histologic pattern analysis of basal cell carcinoma 1125

Table II. Adequacy of excision versus histologic pattern Pattern

Nodular Micronodular Superficial Infiltrative Morpheic Nodular-micronodular Mixed with infiltrative Total

No. of excisions

94 59 56 34 6

115

103 467

togenesis. lO, 11 In our series of 56 specimens of this type, neoplasm was present at the surgical margins in only two. Therefore this neoplasm can be adequately treated by simple excision and/or tissuedestructive techniques such as electrodessication and curettage. Jacobs et a1. 6 also considered the superficial-multifocal-type basal cell carcinoma to be less infiltrative. The infiltrative basal cell carcinoma is associated with a relatively high risk of incomplete removal. Our findings agree with those of Thackray, 4 Sloane, 5 Jacobs et al.,6 and Lang and Maize.? In addition, Siegle et al. 12 reported 29 infiltrative basal cell carcinomas. They described a lack of sharp clinical borders and a larger postoperative defect with recurrent infiltrative basal cell carcinomas versus recurrent nodular neoplasms. In the series by Lang and Maize? of 51 recurrent basal cell carcinomas, the infiltrative type was found as the predominant histologic pattern in 10 (19.6%). Therefore, as also shown by our data, primary infiltrative basal cell carcinomas have a much higher rate of recurrence. Afzelius et a1. B also considered the infiltrating basal cell carcinoma more likely to recur. Recently, Dixon et al. I4 described 30 recurrent basal cell carcinomas and compared them with 74 nonrecurrent neoplasms. Thirty percent of recurrent basal cell carcinomas had an infiltrating pattern versus 13% of nonrecurrent basal cell carcinomas. In addition, they demonstrated that tumors with a spiking configuration were more prone to recurrence. Farmer and Helwig I5 studied 17 cases of metastatic basal cell carcinoma. In one of their cases an infiltrative pattern is depicted. Similarly, Mikhail et al. I6 reported two cases of metastatic basal cell carcinoma. One of their cases demonstrated an infil-

No. of margins Positive (% )

I

Negative

6 11 2 9 2 19

(6.4) (18.6) (3.6) (26.5) (33.3) (16.5) l! (30.1) 80 (17.1%)

88 48 54 25 4 96

72

387 (82.9%)

Table III. Histologic pattern versus ulceration, pigmentation, and cystic change Pattern (total No. of neoplasms)

Ulceration (No.)

Pigmentation (No.)

Cystic change (No.)

Nodular (218) Superficial (181) Micronodular (151) Infiltrative (77) Morpheic (11) Nodular-micronodular

75 31 44 26 2 79

17 13 6 0 0 28

95 1 8 0 0 71

Mixed with infiltrative

82

6

28

(226) (175)

trative basal cell carcinoma and the other case demonstrated a morpheic pattern. However, these authors concluded that overall no specific histologic type of basal cell carcinoma was more capable of metastasizing. Domarus and Stevens!? reported five cases of metastatic basal cell carcinoma in 1984. Four showed an infiltrating pattern and the other neoplasm was nodular. Gormley and Hirsch l8 reported a basal cell carcinoma of the scalp that invaded subcutaneous fat, muscle, fascia, and bone. Photographs of their case showed a micronodular, infiltrative, and morpheic pattern. Mixed pattern basal cell carcinomas behaved similarly to the pattern most likely to result in incomplete remova1. Of 115 excisional specimens of the nodular-micronodular pattern, 16.5% demonstrated tumor at the margin. Of 103 excisions of a mixed pattern with an infiltrative component, 30.1 % demonstrated positive margins. These data indicate that with mixed-type basal cell carcinomas the growth pattern most likely to result in incomplete

1126 Sexton et al. removal is the most important to recognize. The appearance of the pattern in the periphery of the neoplasm is most important. This observation was emphasized by Jacobs et a1. 6 and Dixon et aI.14 The morpheic basal cell carcinoma is uncommon. Jacobs et a1. 6 and Lang and Maize? included one example in each of their series. Salasche and Amonette 19 reported that the average subclinical extension of morpheaform basal cell carcinoma was 7.2 mm (± 3.8 mm). They were unable to define clinically, the borders of this type of neoplasm. Litzow et a1. 20 also emphasized the subtle clinical appearance of morpheic basal cell carcinoma. They concluded that the treatment of choice was Mohs surgery or wide excision with frozen-section control of surgical margins. There are some patterns of basal cell carcinoma (micronodular, infiltrative, and morpheic) in which the chance of residual tumor after conventional excision is high. In these instances, wider margins should be used. Mohs surgery increases the chances of cure. 7, 12,21,22 We also agree with Sloane5 who said that "growth pattern should always be stated in the routine histopathological reporting of these neoplasms" (page 127) so that the correct surgical therapy can be planned and used.

REFERENCES 1. Lever WF, Schaumburg-Lever G. Histopathology of the skin. Philadelphia: JB Lippincott, 1990:622-34. 2. Mehregan AH. Pinkus' guide to dermatohistopathology. Norwalk, Conn: Appleton-Century-Crofts, 1986:509-18. 3. Pascal RR, Hobby LW, Lattes R, et al. Prognosis of "incompletely excised" versus "completely excised" basal cell carcinoma. Plast Reconstr Surg 1968;41:328-32. 4. Thackray AC. Histological classification of rodent ulcers and its bearing on their prognosis. Br J Cancer 1951;5: 213-6.

Journal of the American Academy of Dermatology

5. Sloane JR. The value of typing basal cell carcinomas in predicting recurrence after surgical excision. Br J Dermatol 1977;96:127-33. 6. Jacobs GH, Rippey JJ, Altini M. Prediction of aggressive behavior in basal cell carcinoma. Cancer 1982;49:533-7. 7. Lang PG, Maize J C. Histologic evolution ofrecurrent basal cell carcinoma and treatment implications. J AM ACAD DERMATOL 1986;14:186-96. 8. Tozawa T, Ackerman AB. Basal cell carcinoma with follicular differentiation. Am J Dermatopathol 1987;9:47482. 9. Wermuth BM, Fajardo LF. Metastatic basal cell carcinoma: a review. Arch PathoI1970;90-458-62. 10. Madsden A. The histogenesis of superficial basal-cell epithelioma. Arch DermatoI1955;72:29-30. II. Kimura S. Three-dimensional architecture of epithelial skin tumors: an application of epidermal separation. J Dermato1 (Tokyo) 1981;8:13-9. 12. Siegle RJ, Macmillan J, Pollack SV. Infiltrative basal cell carcinoma: a nonsclerosing subtype. J Dermatol Surg Oncol 1986;12:830-6. 13. Afzelius LE, Ehnhage A, Nordgren H. Basal cell carcinoma in the head and neck. The importance oflocation and histological picture, studied with a new scoring system, in predicting recurrences. Acta Patho1 Microbiol Scand 1980;88:5-9. 14. Dixon AY, Lee SH, McGregor DR. Factors predictive of recurrence of Basal cell carcinomas. Am J Dermatopatho1 1989;11:222-32. 15. Farmer ER, Helwig EB. Metastatic basal cell carcinoma: a clinicopathologic study of 17 cases. Cancer 1980;46:74857. 16. Mikhail GR, Nims LP, Kelly AP Jr, et al. Metastatic basal cell carcinoma: review, pathogenesis and report of two cases. Arch DermatoI1977;113:1261-9. 17. Domarus HV, Stevens PJ. Metastatic basal cell carcinoma. JAM ACAD DERMATOL 1984;10:1043-60. 18. Gormley DE, Hirsch P. Aggressive basal cell carcinoma of the scalp. Arch DermatoI1978;114:782-3. 19. Salasche SJ, Amonnette RA. Morpheaform basal-cell epitheliomas: a study of subclinical extensions in a series of 51 cases. J Dermato1 Surg OncoI1981;7:387-94. 20. Litzow TJ, Perry HO, Soderstrom CWo Morpheaform basal cell carcinoma. Am J Surg 1968;116:499-505.