nevocellular nevi, and solar lentigines. J Invest Dermatol ... Cutaneous melanoma: Clinical management and Treatment Results. Worldwide, 2nd ed. 1991. 13.
Histopathologic Recognition and Grading of Dysplastic Melanocytic Nevi: An Interobserver Agreement Study Lyn M. Duncan, Marianne Berwick, Jan A. Bruijn, H. Randolph Byers, Martin C. Mihm and Raymond L. Barnhill Before the controversies surrounding dysplastic melanocytic nevi are resolved, dermatopathologists must be able to reliably distinguish dysplastic nevi from common acquired nevi and malignant melanoma. To establish whether grading of melanocytic dysplasia has any biologic relevance, dermatopathologists must be able to consistently recognize two or more grades of atypia. We studied the concordance among five dermatopathologists for recognition and grading of 60 nevomelanocytic lesions. Ten cases from each of the following categories of melanocytic proliferation were retrieved from the Massachusetts General Hospital files: (1) common melanocytic nevi, (2) melanocytic nevi with features of dysplastic nevi, (3) dysplastic nevi with slight cytologic atypia, (4) dysplastic nevi with moderate cytologic atypia, (5) dysplastic nevi with severe cytologic atypia, and (6) primary malignant melanoma. The slides were reviewed independently; no discussion of diagnostic criteria preceded the review. Overall concordance for diagnosing dysplastic nevi was 77%, with a kappa statistic of 0.55–0.84. Furthermore, in grading dysplastic nevi, experienced dermatopathologists had a concordance ranging from 35 to 58% (kappa value 0.38–0.47). Those with less experience in grading dysplastic nevi had a concordance of 16–65% (kappa value 0.05–0.24). The five observers in this study reliably distinguished dysplastic nevi from common acquired nevi and malignant melanoma. Further refinement of the criteria for grading of nevomelanocytic dysplasia and experience in grading are critical for accuracy in subcategorization of dysplastic nevi. Consistent, reproducible subcategorization or dysplastic nevi is a requisite before the issue of biologic or prognostic relevance of grading (dysplastic nevi) can be addressed. This study supports the validity of existing criteria for the diagnosis of dysplastic nevi because the problems in diagnosis were at the limits of the spectrum, namely, discrimination of slightly atypical dysplastic nevi from common nevi and severely atypical dysplastic nevi from radial growth phase melanoma. J Invest Dermatol 100:318S–321S, 1993
Recent studies suggest that melanocytic dysplasia is an intermediate step between common acquired nevi and melanoma [1–7]. If recognition of dysplastic melanocytic nevi has biologic significance, then pathologists must be able to consistently distinguish dysplastic nevi from melanoma and common nevi. Likewise, if melanoma risk is related to degree of cytologic atypia in dysplastic nevi, dermatopathologists must be able to reliably distinguish two or more grades of melanocyte atypia in dysplastic nevi. Histologic criteria for dysplastic nevi are not standardized at present [1]. Most observers emphasize a variable constellation of architectural features in association with cytologic atypia of intraepidermal nevomelanocytes [8,9]; however, the need for cytologic atypia has been dismissed by some pathologists because such atypia is often subtle and not reproducible. Prior to this report, two studies showed that dysplastic nevi may be diagnosed with Division of Dermatopathology (LMD, HRB, MCM, RLB), Massachusetts General Hospital, Boston, Massachusetts; Cancer Prevention Research Institute (MB), New York, New York, U.S.A.; and Laboratorium voor Pathologie (JAB), Leiden, The Netherlands Reprint requests to: Dr. Raymond L. Barnhill, Department of Pathology, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115. Abbreviations: MGH, Massachusetts General Hospital; WHO, World Health Organization
0022-202X/93/$06.00
some consistency by pathologists [8,10,11]. Although criteria for cytologicatypia are not well established, attempts have been made to define grades of melanocyte atypia [11]. We examined concordance and interobserver agreement for the recognition and grading of 60 nevomelanocytic lesions. The purpose of the study was to assess the agreement among five observers in (1) distinguishing dysplastic nevi from melanoma and common acquired nevi, and (2) grading of dysplastic nevi.
METHODS Sixty melanocytic lesions were randomly retrieved from the dermatopathology files at Massachusetts General Hospital (MGH) on the basis of the final reported diagnoses from the previous 6 months. These 60 cases were composed of 10 lesions from each of the following diagnostic categories: (1) common acquired melanocytic nevi (limited to lentiginous junctional and lentiginous compound nevi, three of which had some features of congenital onset); (2) melanocytic nevi with features of dysplastic nevi (one of which also had some features of congenital onset); (3) dysplastic nevi with slight cytologic atypia; (4) dysplastic nevi with moderate cytologic atypia; (5) dysplastic nevi with severe cytologic atypia; and (6) primary malignant melanoma in the radial growth phase. Criteria for diagnosing dysplastic nevi and grading of atypia were based
Copyright & 1993 by The Society for Investigative Dermatology, Inc. 318S
VOL.
100, NO. 3, SUPPLEMENT, MARCH 1993
RECOGNITION AND GRADING OF DYSPLASTIC NEVI
on those previously published [7,11,12]. A single representative slide was chosen from each case by one of the observers (RLB). The slides were evaluated by the five observers independently. Each participant was provided with a flowsheet on which they were requested to check a single diagnosis (of the six listed above) for each case. If a dysplastic nevus was diagnosed, the observer was requested to grade the degree of cytologic atypia as slight, moderate, or severe. There was no discussion of diagnostic criteria prior to the review. Three of the observers were staff dermatopathologists at MGH with experience ranging from 6 years to more than 20 years. Two of the observers were dermatopathologists in training, both pathologists, but each with less than 2 months experience at an institution where grading of dysplastic nevi is routinely performed. The data were analyzed for concordance and by the kappa statistic. Concordance was evaluated by Kendall’s taub using the Corr procedure [13]. This is a measure calculated from values of paired observations as to whether they vary together (in concord) or differently (in discord). The kappa statistic allows two or more observers to be compared, with an allowance made for chance. Originally described by Cohen [14], the higher the kappa value, the better the agreement. Kappa values greater than 0.75 indicate excellent agreement beyond chance; less than 0.40 indicate poor agreement beyond chance; and 0.40–0.75 indicate fair to good agreement beyond chance. Only data with a sensitivity of less than 0.05 are presented.
RESULTS The distribution of diagnoses by each observer and the final reported diagnoses are shown in Table I. Because the cases were randomly selected from the MGH files, there was no selection against difficult cases. Two cases included posed diagnostic dilemmas. In fact, each staff member diagnosed these differently. One staff member (A) called both lesions malignant melanoma; another observer (B) diagnosed both lesions as dysplastic nevi with severe cytologic atypia; and the third member of the staff (C) called one lesion a melanoma and the other a dysplastic nevus with severe cytologic atypia. Nonetheless, the concordance among the staff for diagnosing melanoma was 90% with a kappa value ranging from 0.84 to 0.94 (These data are calculated on the ability of observers to distinguish radial growth phase melanoma from common acquired nevi, nevi with features of dysplastic nevi, and all three grades of dysplastic nevi.). The trainees diagnosed melanoma with a concordance of 88% and a kappa value ranging from 0.83 to 1.0. The overall concordance for discriminating melanoma from all nevi, including dysplastic nevi, was 90% (Table II).
Diagnosis
Final
A
B
C
D
E
Benign nevi (including those with 20 features of dysplasia)
20
23
20
29
21
23
Dysplastic nevi (total)
30
25
30
20
28
27
Slight atypia
10
14
11
12
6
6
Moderate atypia
10
5
10
4
15
14
Severe atypia Malignant melanoma a
The statistics for diagnosing dysplastic nevi are shown in Table II and Table III. These data show the concordance and interobserver agreement in distinguishing dysplastic nevi from ‘‘nevi’’ on one end of the spectrum and melanoma on the other end. ‘‘Nevi’’ in these tabulations include common nevi and nevi with architectural features of dysplasia or cytologic atypia but with insufficient criteria for the diagnosis of dysplastic nevus. The overall concordance in diagnosing dysplastic nevi was 77% with a kappa statistic of 0.55–0.84. The staff diagnosed dysplastic nevi with a concordance of 69–80% and a kappa statistic of 0.63–0.71, whereas the trainees showed concordance of 61–88% with a kappa statistic of 0.55–0.84. Data for trainees include their agreement with each other, as well as their agreement with staff members. The concordance and kappa statistic for each observer pair are shown in Table III. The data for grading dysplastic nevi are also shown in Table II and Table III. The overall concordance was 17–65% among observers and 36–65% with the reported diagnosis (data not shown). The overall kappa values ranged from 0.05 to 0.47. The staff dermatopathologists were able to grade dysplastic nevi with a concordance ranging from 35 to 58% and kappa values ranging
Table II. Agreement in Diagnosis of Melanocytic Proliferationsa Overall
Staff
Trainees
All nevi (including dysplastic nevi) versus melanoma
90% (0.83–1.0)
84–94% (0.84–0.94)
83–100% (0.83–1.0)
Nevi versus dysplastic nevi versus melanoma
77% (0.55–0.84)
69–80% (0.63–0.71)
61–88% (0.55–0.84)
Dysplastic nevi Slight versus moderate versus severe cytologic atypia
17–65% (0.05–0.47)
35–58% (0.38–0.47)
17–65% (0.05–0.30)
a
Data for concordance are expressed as percentages. The range of kappa statistic is listed below in parentheses.
Table III. Interobserver Agreementa Nevi Versus Dysplastic Nevi Versus Melanoma Observer Pairs
Table I. Diagnosis by Observera
10
6
9
4
7
7
10
12
10
11
11
10
Number of cases diagnosed in each category by observers A-E; ‘‘Final’’ designates the reported diagnosis. Benign nevi include common nevi and nevi with some features, either cytologic or architectural, of dysplastic nevi but not fulfilling all of the criteria.
319S
Slight Versus Moderate Versus Severe
Concordance Cohen’s Kappa Concordance Cohen’s Kappa (%) Value (%) Value
A versus B
80
0.71
58
0.47
A versus C
69
0.63
35
0.44
B versus C
78
0.67
45
0.38
A versus D
72
0.61
39
0.24
A versus E
62
0.55
17
0.20
B versus D
88
0.84
65
0.23
B versus E
80
0.75
28
0.24
C versus D
74
0.64
38
0.05
C versus E
78
0.69
46
0.13
D versus E
78
0.71
46
0.30
a
The first three rows show agreement among staff dermatopathologists.
320S
DUNCAN ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
from 0.38 to 0.47 (i.e., showing moderate agreement beyond chance). In contrast, the trainees graded dysplastic nevi with a concordance of 16–65% and kappa values ranging from 0.05 to 0.30, indicating poor to moderate agreement beyond chance. The interobserver agreement for diagnosing severely atypical dysplastic nevi versus melanoma ranged from 0.41 to 1.0 (staff 0.41–0.81). On the other hand, interobserver agreement for diagnosis of slightly atypical dysplastic nevi versus nevi with features of dysplasia ranged from 0.19 to 0.75 (staff 0.40–0.59).
DISCUSSION Dysplastic nevi have been the topic of much debate and controversy. Some observers have objected to the term ‘‘dysplasia’’ as applied to nevi and the ‘‘B-K mole’’ or dysplastic nevus syndrome and have disputed the concept that dysplastic nevi are common precursors of melanoma [15–17]. Some have also suggested that many of the histologic changes noted in dysplastic nevi lack specificity and that melanocytic nevi with such changes may have a prevalence as high as 53% in the general population [18]. As a result of the pioneering work of Clark and co-workers [19,20], Lynch et al [21,22], and others [23,26], it is now accepted that people with increased numbers of clinically atypical (dysplastic) nevi are at an increased risk for developing malignant melanoma, particularly in the context of hereditary melanoma. Although most workers have defined the dysplastic nevus using both clinical and pathologic criteria [27,28], the clinical phenotype of numerous, often grossly abnormal, nevi appears to be the primary risk factor for melanoma [26]. Partly because of the lack of agreement on the histopathologic definition of dysplastic nevi, the association of melanoma risk with histoligic changes of dysplastic nevi has not been clearly established [1]. The lack of correspondence between clinically dysplastic nevi and histologically dysplastic nevi has been well documented [18,29,30]. This is not surprising, given the lack of universally accepted criteria for dysplastic nevi. Most notably, many such studies did not require cytologic atypia for the histologic diagnosis of dysplastic nevi. This was in part a result of the first consensus conference on precursor lesions of melanoma, which did not require melanocytic atypia for the diagnosis of dysplastic nevi [31]; however, most observers currently emphasize that melanocytic atypia is an obligatory histopathologic criterion for the diagnosis of dysplastic nevi [1,8]. The histopathologic criteria used in this study to diagnose dysplastic nevi have been previously described [1–9,11,12]. Both architectural and cytologic atypia are required for the diagnosis of dysplastic nevus; however, only limited quantitative data are available concerning histopathologic criteria for dysplastic nevi. Barnhill et al [7] studied the histologic characteristics of the clinically most atypical nevus removed from each of 153 patients with non-familial malignant melanoma. Assuming that cytologic atypia of intraepidermal nevomelanocytes is mandatory for recognition of dysplastic melanocytic nevi, Barnhill et al identified nevi with unequivocal cytologic atypia of the latter population of cells based on two independent, temporally distinct reviews. Subsequently, 16 architectural features ascribed to dysplastic nevi were correlated with cytologic atypia by univariate and multivariate analysis in this series of nevi. On the basis of results of multivariate analysis, Barnhill et al concluded that abnormal patterns of intraepidermal melanocytic
proliferation, either lentiginous, irregular junctional nesting, or both, are essential features of dysplastic nevi in addition to cytologic atypia. Architectural features of melanocytic dysplasia emphasized by the World Health Organization (WHO) consensus report included (1) basilar nevomelanocytic proliferation (extending at least three rete ridges beyond any dermal nevocellular component); (2) lamellar fibrosis or concentric eosinophilic fibrosis; (3) neovascularization; (4) inflammatory response; and (5) fusion of rete ridges [8]. Criteria for cytologic atypia in dysplastic nevi have included nuclear enlargement, pleomorphism, hyperchromatism, and prominent nucleoli [7,11,12]. The five dermatopathologists in the present study reliably distinguished dysplastic nevi from melanoma with a mean concordance of 90%. This is remarkably similar to the 92% concordance achieved by the pathologists in the WHO melanoma program [8]. The WHO study, however, also included distinction of dysplastic nevi from common nevi. In the present study, the dermatopathologists distinguished dysplastic nevi from common nevi (including those with only some features of dysplasia) and melanoma with a concordance of 77%. Hence, without prior discussion of histologic criteria, these observers reliably made the diagnosis of dysplastic nevus with good to excellent interobserver agreement beyond chance (kappa value 0.55–0.84). Observers were more reliable in distinguishing severely atypical dysplastic nevi from melanoma (kappa value 0.41–1.0) than in distinguishing slightly atypical dysplastic nevi from nevi with features of dysplasia (kappa value 0.19–0.75). The greater difficulty in making the latter discrimination indicates that recognition of subtle architectural and cytologic alterations is problematic and possibly not reproducible. In our opinion, there is a morphologic distinction between dysplastic nevus with severe cytologic atypia and early radial growth phase malignant melanoma; however, in some cases this distinction may be impossible to make because these entities represent two close points along a spectrum of morphologic changes. The importance of grading dysplastic nevi rests on the notion that the biologic behavior of nevi at either end of the spectrum of dysplasia is different and that dysplastic nevi represent the middle ground in a continuum from benign common nevi to malignant melanoma. This is supported by evidence that dysplastic nevi show both clinical and histopathologic features intermediate between those of truly benign nevi and overtly malignant melanomas. Accumulating data from other sources, such as immunophenotyping with cell surface markers and cytogenetics, support this idea as well [1,32–34]. The advantage of a three-grade scale over a two-grade scale may be the allowance of a middle ground for cases where the degree of atypia is not obviously slight or severe. In fact, both of the less experienced observers appeared to use this ‘‘safe haven’’ and graded a disproportionate number of dysplastic nevi as moderate (Table I). The data for grading were also affected by the lack of discussion of criteria prior to review of the slides. The terms ‘‘slight to focal moderate’’ and ‘‘moderate to focal severe’’ are used in daily practice by the authors but were not discussed prior to review. Hence, observers were forced to place the dysplastic nevi into a single category of ‘‘slight,’’ ‘‘moderate,’’ or ‘‘severe’’ cytologic atypia. This procedure almost certainly resulted in less agreement for particular lesions. Consistency in subcategorization of dysplastic nevi is necessary before the issue of biologic and prognostic relevance of
VOL.
100, NO. 3, SUPPLEMENT, MARCH 1993
grading dysplastic nevi can be addressed. First, further refinement of the criteria for grading of nevomelanocytic dysplasia is needed. This may include correlation with DNA content, nuclear volume, and other objective parameters [32–35]. Second, experience in grading nevomelanocytic dysplasia is necessary. In the present report, we have shown that even dermatopathologists in training are able to recognize melanocyte atypia (as is required for the diagnosis of dysplastic nevi). Dermatopathologists in training were able to distinguish dysplastic nevi from common acquired nevi and melanoma with a concordance ranging from 61 to 88% and kappa statistic ranging from 0.55 to 0.84. We also found that (1) grading melanocytic atypia may be learned (staff were able to grade with greater agreement than trainees), and (2) greater communication among histopathologists about criteria and more objective criteria of melanocyte atypia are needed (there was only moderate agreement in grading among the staff). On the other hand, grading of atypia is inherently subjective, and there are limits to the ability to make such subtle distinctions on a consistent basis. This study also supports the validity of existing criteria for the diagnosis of dysplastic nevi because the problems in diagnosis were at the limits of the spectrum, namely, discrimination of slightly atypical dysplastic nevi from common nevi and severely atypical dysplastic nevi from melanoma. Defining the limits of such a spectrum is difficult and often arbitrary; however, this does not negate the value of establishing histologic criteria for the diagnosis and grading of dysplastic nevi.
REFERENCES 1. Barnhill RL: Current status of the dysplastic melanocytic nevus. J Cutan Pathol 18:147–159, 1991 2. Barnhill RL, Hurwitz S, Duray PH, Arons MS: The dysplastic nevus: recognition and management. Plast Reconstr Surg 81:280–289, 1988 3. Clark WH Jr, Elder DE, Guerry DIV, et al A study of tumor progression: the precursor lesions of superficial spreading and nodular melanoma. Hum Pathol 15:1147–1165, 1984 4. Rhodes AR, Melski JW, Sober AJ, Harrist TJ, Mihm MC Jr, Fitzpatrick TB: Increased intraepidermal melanocyte frequency and size in dysplastic melanocytic nevi and cutaneous melanoma. A comparative quantitative study of dysplastic melanocytic nevi, superficial spreading melanoma, nevocellular nevi, and solar lentigines. J Invest Dermatol 80:452–459, 1983
RECOGNITION AND GRADING OF DYSPLASTIC NEVI
321S
11. Rhodes AR, Mihm MC Jr, Weinstock MA: Dysplastic melanocytic nevi: a reproducible histologic definition emphasizing cellular morphology. Mod Pathol 2:306–319,1989 12. Barnhill RL, Mihm MC Jr: Histopathology of malignant melanoma and its precursors. In: Balch CM, Houghton A, Milton G, Sober A (eds.). Cutaneous melanoma: Clinical management and Treatment Results Worldwide, 2nd ed. 1991 13. Sas Institute Inc., Cary, NC, Version 6.03, 1988 14. Cohen J: Weighted kappa: nominal scale agreement with provision for scaled disagreement or partial credit. Psychol Bull 70:213, 1968 15. Ackerman AB, Mihara I: Dysplasia, dysplastic melanocytes, dysplastic nevi, the dysplastic nevus syndrome, and the relation between dysplastic nevi and malignant melanomas. Hum Pathol 16:87–91, 1985 16. Ackerman AB: What naevus is dysplastic, a syndrome and the commonest precursor of malignant melanoma? A riddle and an answer. Histopathology 13:156–241, 1988 17. Briggs JC: Invited review: melanoma precursor lesions and borderline melanomas. Histopathology 9:1251–1262, 1985 18. Piepkorn M, Meyer LJ, Goldgar D, et al: The dysplastic melanocytic nevus. A prevalent lesion that correlates poorly with clinical phenotype. J Am Acad Dermatol 20:407–415, 1989 19. Clark WH Jr, Reimer RR, Greene MH, et al: Origin of familial malignant melanomas from heritable melanocytic lesions. The BK mole syndrome. Arch Dermatol 114:732–738, 1978 20. Reimer RR, Clark WH Jr, Greene MH, et ah Precursor lesions in familial melanoma. A new genetic preneoplastic syndrome. JAMA 239:746–774, 1978 21. Lynch HT, Frichot B III, Lynch JF: Familial atypical multi-mole melanoma syndrome. J Med Genet 15:352–356, 1978 22. Frichot BC, Lynch HT, Guirgis HA, et al New cutaneous phenotype in familial malignant melanoma. Lancet 1:864, 1977 23. Greene MH, Clark WH, Tucker MA, et al High risk of malignant melanoma in melanoma-prone families with dysplastic nevi. Ann Intern Med 102:458–465, 1985 24. Rigel DS, Rivers JK, Kopf AW, Friedman RJ, Vinokur AF, Heilman ER, Levenstein M: Dysplastic nevi: markers for increased risk for melanoma. Cancer 63:386–389, 1989 25. Albert LS, Rhodes AR, Sober AJ: Dysplastic melanocytic nevi and cutaneous melanoma. Markers of increased melanoma risk for affected persons and blood relatives. J Am Acad Dermatol 22(1):69–75, 1990 26. Holly EA, Kelly JW, Shpall SN, Chiu S-H: Number of melanocytic nevi as a major risk factor for malignant melanoma. J Am Acad Dermatol 17:459, 1987 27. Elder DE, Goldman LI, Goldman SC, Greene MH, Clark WH Jr: The dysplastic nevus syndrome: a phenotypic association of sporadic cutaneous melanoma. Cancer 46:1787–1794, 1980
5. Elder DE, Clark WH Jr: Developmental biology of malignant melanoma. Pigm Cell 8:1–28, 1987
28. Elder DE, Greene MH, Guerry DIV, Kraemer KH, Clark WH Jr: The dysplastic nevus syndrome: our definition. Am J Dermatopathol 4:455–460, 1982
6. Tong AKF, Murphy GF, Mihm MC Jr: Dysplastic nevus: a formal histogenetic precursor of malignant melanoma. In: Mihm MC Jr, Murphy GF, Kaufman N (eds.). Pathology and Recognition of Malignant Melanoma. Williams & Wilkins, Baltimore, 1988, pp 10–18
29. Fitzpatrick TB, Rhodes AR, Sober AJ: Prevention of melanoma by recognition of its precursors. N Engl J Med 312:115, 1985
7. Barnhill RL, Roush GC, Duray PH: Correlation of histologic architectural and cytoplasmic features with nuclear atypia in atypical (dysplastic) nevomelanocytic nevi. Hum Pathol 21:51–58, 1990 8. Clemente C, Cochran AJ, Elder DE, Levene A, MacKie RM, Mihm MC Jr, Rilke F, Cascinelli N, Fitzpatrick TB, Sober AJ: Histopathologic diagnosis of dysplastic nevi: concordance among pathologists convened by the World Health Organization Melanoma Programme. Hum Pathol 22:313–319, 1991 9. Roush GC, Barnhill RL, Duray PH, Titus LJ, Ernstoff MS, Kirkwood JM: Diagnosis of dysplastic nevus in different populations. J Am Acad Dermatol 14:419–425, 1986 10. Duray PH, DerSimonian R, Barnhill RL, Stenn K, Ernstoff MS, Kirkwood HM, Fine J: Interobserver recognition of dysplastic nevi from a mixed group. J Cutan Pathol 17:291, 1990
30. Grob JJ, Andrac L, Ramano MH, Davin D, Collet-Villette AM, Munoz MH, Bonerandi JJ: Dysplastic naevus in non-familial melanoma. A clinicopathological study of 101 cases. Br J Dermatol 118:745–752, 1988 31. National Institutes of Health Consensus Conference. Precursors to malignant melanoma. JAMA 251:1864–1866, 1984 32. Bergman W, Ruiter DJ, Scheffer E, et al Melanocytic atypia in dysplastic nevi. Immunohistochemical and cytophotometrical analysis. Cancer 61:1660–1668, 1988 33. Parmiter AH, Balaban G, Clark WH Jr, Nowell PC: Possible involvement of the chromosome region 10q24-q26 in early stages of melanocytic neoplasia. Cancer Genet Cytogenet 30:313, 1988 34. Cowan JM, Halaban R, Francke U: Cytogenetic analysis of melanocytes from premalignant nevi and melanomas. J Natl Cancer Inst 80:1159,1988 35. Schmiegelow P, Schroiff R, Breitbart E, et al: Malignant melanoma – its precursor and topography of proliferation: DNA-Feulgen cytophotometry and mitotic index. Virchows Arch (A) 409:47–55, 1986
