A Study of Tumor Progression: The Precursor Lesions of Superficial Spreading and Nodular Melanoma. Clark WH Jr, Elder DE, Guerry D IV, Epstein MN, Greene MH, and Van Horn M. Human Pathology 1984;15:1147–65.

"Six evident lesions steps of tumor progression form the neoplastic system that affects the human epidermal melanocyte: 1) the common acquired melanocytic nevus; 2) a melanocytic nevus with lentiginous melanocytic hyperplasia, i.e., aberrant differentiation; 3) a melanocytic nevus with aberrant differentiation and melancytic nuclear atypia, i.e., melanocytic dysplasia; 4) the radial growth phase of primary melanoma; 5) the vertical growth phase of primary melanoma; and 6) metastatic melanoma. The common acquired melanocytic nevus is viewed as a focal proliferation of melanocytes, destined in most instances to follow a programmed pathway of differentiation that leads to disappearance of the nevus. If the pathway of differentiation is not followed, characteristic lesions result, and such lesions are regarded as the formal histogenetic precursors of melanoma. Such a developmental flaw is termed aberrant differentiation, and the resultant precursor lesion is designated melanocytic dysplasia. The vast majority of melanocytic nevi showing melanocytic dysplasia are terminal lesions that do not progress to melanoma. If melanoma is to develop via a precursor lesion, however, the nevus with melanocytic dysplasia is that precursor. When melanomas do develop, they develop focally within the precursor. The resultant primary melanoma itself does not follow a pathway of inexorable expansion of a population of melanoma cells in space and time. Rather, primary melanomas, with the exception of nodular melanoma, also evolve in a stepwise fashion.

The lesions of tumor progression described in this paper are thought to be paradigm for neoplasia, and from this model a sequence of generic lesions applicable to neoplastic development in general is presented. These generic steps of tumor progression are 1) a selective focal proliferation of structurally normal cells (a benign tumor); 2) an abnormal pattern of hyperplasia (aberrant differentiation); 3) an abnormal pattern of hyperplasia and random cytologic atypia (aberrant differentiation and the appearance of cells with nuclear atypia); 4) primary cancer without competence for metastasis; 5) primary cancer with competence for metastasis; and 6) metastatic cancer."

In December 1984, Clark and co-workers published their longest article to date about dysplasia, DN, and the DNS (Fig. 17) In it the authors elaborated on six steps that they conceived to be essential to melanocytic tumor progression as follows:

The progression from step 1 to 2, and from 2 to step 3 was predicated on a misperception of Richard Reed, namely, that "dysplastic nevi differ from common acquired nevi" by virtue of "shoulders" i.e., a junctional component that extends beyond the intradermal one. At first, Reed and co-workers, such as Clark and Elder, referred to those junctional changes as "atypical melanocytic hyperplasia," then as "aberrant differentiation," and last as "dysplasia." By incorporating those expected findings at the shoulders of an extraordinarily common type of melanocytic nevus, which they dubbed DN, into their concept of tumor progression," Clark et al. attempted to explain how some nevi, which they called CAN, became dysplastic and then progressed to melanoma. The notion of "tumor progression" advanced in this article in Human Pathology formed the basis for the claim by Clark and co-workers of legitimacy of their concept of DNS.

Clark et al.: "Primary melanoma does not simply progress uniformly in space and time; rather, its evolution is stepwise. Each step, in both the precursor lesions and in established melanomas, is characterized by the acquisition of properties not manifested in the preceding lesions. In this paper we treat the early developmental steps of primary superficial spreading melanoma to distinguish them from the steps that precede them. A companion paper (W.H. Clark, Jr., D.E. Elder, D. Guerry, IV, et al., in preparation) will fully describe tumor progression occurring within established melanomas."

Critique: There is no compelling evidence that cutaneous melanoma evolves stepwise through a series of a few discrete, definable "steps"; it may be a continuous process. The same may be said of the evolution of squamous-cell carcinoma of the cervix. So-called dysplasia of the cervix is a hodge-podge of inflammatory and neoplastic processes. Reputed "precursor lesions" of cervical carcinoma often are designated dysplasia when, in actuality, the changes are simply those of evolving carcinoma. The same is true for cutaneous squamous-cell carcinomas known as solar keratoses and Bowen's disease, and for cutaneous melanoma. Many of the so-called precursor lesions in the epidermis in conjunction with cutaneous melanoma, referred to by Clark et al. as dysplasia, are, in fact, melanoma entirely, i.e., melanoma in situ. In other instances, so-called dysplasia is merely the shoulders of a Clark's nevus, i.e., a junctional component invariable in compound Clark's nevi. In short, there is no true "melanocytic dysplasia" intermediate between junctional nevus and melanoma in situ. The diagnoses of junctional nevus and of melanoma in situ usually can be made with confidence by microscopy if criteria now available are employed scrupulously.

Clark et al.: "The first evident focal proliferation of melanocytes is the common acquired melanocytic nevus (nevocellular nevus, melanocytic nevus, common nevus, nevus)."

Critique: Clark et al. write about CAN as if it is a single specific type of melanocytic nevus when, it actuality, there are several very different kinds of CAN, e.g., Spitz's (various configurations formed by melanocytes with large nuclei, abundant cytoplasm, and round, plump, oval, and polygonal shapes especially) and Unna's type (configuration of a fibroepithelial polyp and small, oval monomorphous melanocytes confined to a tremendously thickened papillary dermis), in addition to the most common of all acquired nevi, i.e., Clark's type. It is the Clark's type of common acquired nevus to which the authors allude when referring to the nevus at the bottom rung of their ladder of tumor progression. The shoulders of Clark's type of common acquired nevus already are present before any neoplastic melanocytes in the center of the nevus enter the papillary dermis, but they only become recognizable as shoulders after neoplastic melanocytes have descended into the papillary dermis. Clark and colleagues, in contrast, conceive of the "dysplastic" changes at the shoulders having come into being only after a compound CAN has been "stable" for some years. In reality, the intradermal component of that nevus simply is not as broad as the original junctional component of the nevus.

Clark et al.: "Lesions of the face and scalp, especially the latter, commonly evolve more rapidly (first and second decades of life) than do truncal nevi into globoid, frequently pigment-free lesions with extensive neurotization."

Critique: The dome-shaped melanocytic nevus of the face and scalp just referred to qualifies as what we call Meischer's nevus, a lesion prominent to the visage of such historical figures as Lincoln and Khrushchev. Although it is an acquired nevus in a temporal sense, it has histopathologic attributes of a congenital nevus, namely, involvement of much of the reticular dermis (and sometimes the uppermost part of the subcutaneous fat) by melanocytes splayed between bundles of collagen. In brief, Miescher's nevus is a dome-shaped congenital nevus of tardive type, akin to some lesions of nevus of Ota that, although congenital, do not become apparent until puberty. It is very different from Clark's nevus clinically, histopathologically, and biologically.

Clark et al.: "It has been stated that the peripheral extent of growth of intraepidermal melanocytes defines the lateral margins of the dermal component of nevi and that the nesting of melanocytes within the epidermis, in preparation for migration into the dermis, is apparently the first manifestation of the pathway of differentiation (R.J. Reed, unpublished data)."

Critique: All of the work of Clark and co-workers about melanocytic dysplasia, DN, and DNS is predicated on the "unpublished data" of Richard J. Reed captured in this single sentence. Reed contended that "normal" CAN do not have shoulders, i.e., extension of nests of intraepidermal melanocytes beyond the dermal component of the nevus. Because compound DN invariably display shoulders, Clark, Elder, other collaborators of Reed, and Reed himself regarded the shoulders of the commonest of all melanocytic nevi as abnormal or aberrant, and as premalignant dysplastic changes that were superimposed on a pre-existing CAN. Because this unpublished observation of Reed is not consonant with the actual situation in DN (all compound examples display shoulders, in contrast to Miescher's nevus which actually served as Reed's model), everything that derived from it in regard to DN and the DNS is flawed.

In our assessment, shoulders in Clark's nevi are integral to them, as much expected in them as is confinement of the intradermal component of those nevi to the papillary dermis.¹² Had Reed used Clark's nevus as his model, he would have come to an entirely different conclusion about the prevalence of shoulders, namely, that compound Clark's nevi exhibit shoulders invariably. As a consequence of Reed's misperception about the absence of shoulders from Clark's nevi, Clark and co-workers were led down a cul-de-sac. When, in 1976 and 1977, they biopsied only "funny-looking moles" in patients with familial melanoma, they were stuck by the presence, consistently, of shoulders in those nevi. They then extended this highly subjective image to those nevi histopathologically, i.e., they were considered to be funny looking because of their shoulders. At first the co-workers thought that the changes at the shoulders were those of melanoma in situ but soon came to regard those histopathologic findings as "atypical melanocytic hyperplasia," then as "aberrant differentiation with melanocytic nuclear atypia" and last as just "dysplasia." In truth, those findings were merely the junctional component of the nevus at its periphery—an expected change and nothing more. When Clark and associates began to biopsy B-K moles in 1976, they did not biopsy the lesions that they regarded as CAN. Had they done that they would have discovered that the so-called CAN also exhibited shoulders like those of B-K moles; CAN on the trunk as shown in all of the photographs published by Clark et al. are simply the usual presentation of Clark's nevi. "Funny looking moles" are the unusual clinical presentation of Clark's nevi and are no more "dysplastic" histopathologically than is the usual clinical presentation of these nevi.

Clark et al.: "Nevi in some patients do not follow either of the two evolutionary pathways described (lesional persistence with stability or differentiation). In such nevi peripheral growth of intraepidermal melanocytes, rather than ceasing, persists. This persistent growth forms a clinically apparent, irregular, tan-brown macular component at the periphery of the lesion . . . The growth of melanocytes is basilar (Fig. 17) . . . Nesting of melanocytes may be absent, and such lentiginous growth is similar to that seen in common acquired melanocytic nevi as they first emerge. The usual pattern of differentiation seen in the common melanocytic nevus is no longer followed. The phenomenon of continued intraepidermal growth of melanocytes at the shoulder of a normal nevus may thus be termed aberrant differentiation: the normal nevic pathway of differentiation is flawed."

Critique: The concept of "aberrant differentiation" is based on the notion that "the normal nevic pathway of differentiation is flawed," but the concept of "normal nevic pathway" itself is flawed. There are no "normal" nevi and there is no "normal nevic pathway." There are various types of acquired melanocytic nevi, all of them are abnormal (pathologic), and each type evolves differently, e.g., Unna's, Spitz's, Clark's. Tumor progression, according to Clark et al., proceeds stepwise from shoulders appearing in CAN, considered by them to be stable lesions, a phenomenon that they designate "aberrant differentiation." What they perceived to be aberrant differentiation of DN, however, was simply an expected finding in the most common of all nevi.

Clark et al.: "Cytologically, atypical melanocytes may appear in the area of persistent melanocytic growth at the shoulder of a nevus (aberrant differentiation). Such atypical cells vary from one nevus to another, but two forms are apparent. The first is seen within a prominent area of lentiginous melanocytic hyperplasia. Characteristically, it appears as a large, hyperchromatic nucleus surrounded by a rather sparse amount of cytoplasm, frequently showing artifactual shrinkage. We have described this type of atypia as lentiginous melanocytic dysplasia (Figs. 18 and 19). (Our Fig. 19)"

Critique: Clark et al. termed the third step in the progression of common nevi to melanoma "melanocytic nevi with persistent lentiginous melanocytic hyperplasia (aberrant differentiation) and atypical melanocytic hyperplasia: melanocytic dysplasia." The supposition that hyperplasia can be persistent, however, does violence to fundamental principles of general pathology. By definition, hyperplasia is reversible, in contrast to neoplasia which, as a rule, is not. So-called persistent lentiginous melanocytic hyperplasia is neoplasia. Furthermore, the notion that atypical melanocytes in a zone of lentiginous melanocytic hyperplasia qualifies that change as lentiginous melanocytic dysplasia has no basis in established principles of general pathology. Parenthetically, hardly ever have we seen indubitably atypical melanocytes anywhere in nevi that Clark et al. call dysplastic. Clark et al. consider the melanocytes to be atypical because they possess "large, hyperchromatic" nuclei, but they state nothing about pleomorphism of nuclei, the most important sign of cytologic atypia. Their photomicrographs of lentiginous melanocytic hyperplasia in Fig. 19 (their Figs. 18–19) do not show nuclear atypia. Curiously, as already has been mentioned, some Spitz's nevi display striking nuclear atypia, but the proponents of DN do not refer to them as "dysplastic."

Clark et al.: "The second form of atypical melanocyte is larger, owing to an abundance of cytoplasm, and usually contains finely divided pigment. This cytoplasm rarely shows artifactual shrinkage and surrounds a large nucleus that tends to be spherical and somewhat less chromatic than those of lentiginous melanocytic atypia. We have termed this second type of atypia epithelioid melanocytic dysplasia . . . These atypical cells may be mixed with areas of lentiginous melanocytic atypia, or they may be present as isolated cells at the shoulder of a nevus or in the epidermis over the central region of a nevus."

Critique: Abundance of cytoplasm is irrelevant to a judgment about whether a melanocyte is atypical. Cytologic atypia in classic pathology is based entirely upon nuclear characteristics. The melanocytes that Clark et al. claim to be atypical show small monomorphous nuclei that are not hyperchromatic; they are are into atypical. Lentiginous and epithelioid melanocytic hyperplasia are not contrasting ("lentiginous" hyperplasia refers to a pattern of an increased number of melanocytes, whereas "epithelioid" hyperplasia turns on cytologic features). Neither is a hyperplasia; both are expressions of a benign neoplasm (an acquired melanocytic nevus). Neither is associated with nuclear atypia of melanocytes. These liberties taken by Clark and associates with classic concepts of hyperplasia, neoplasia, and dysplasia, thwart understanding of those subjects by serious students of them. A language as garbled as pig Latin was hailed universally by general pathologists as an analogue of the Latin of Cicero. They made believe that they understood pig Latin perfectly, and soon they began to speak and write it, too.

Clark et al.: "The combination of persistent lentiginous melanocytic hyperplasia (aberrant differentiation) and melanocytic nuclear atypia constitutes melanocytic dysplasia."

Critique: This definition of melanocytic dysplasia by Clark and co-workers continued to be espoused by them despite the unambiguous statement by the NIH Consensus Conference Panel of 1983 that melanocytic nuclear atypia is "often present but not essential for the diagnosis" of dysplastic nevi.14 The combination of "aberrant differentiation" and "melanocytic nuclear atypia" represents the fifth entirely different definition of dysplasia published by Clark et al. between 1977 and 1984.

Clark et al.: "The atypical cells of both lentiginous melanocytic dysplasia and epithelioid melanocytic dysplasia are not uniform and are usually rather widely separated. This separation and variability have led us to term the generic step in tumor progression that involves such changes random cytologic atypia. The atypical cells of melanocytic dysplasia may grow in cluster of three to four cells, forming small nests, but such clusters, if present, are also widely separated from each other. Such nests frequently bud from the epidermis, also in a random fashion, emerging from the sides of rete or over dermal papillae, unlike normal melanocytic nests, which are typically at the tips of rete."

Critique: Random cytologic atypia, as conceived by Clark et al., is unique in general pathology, pertaining only to the nevus they consider to be dysplastic. For them, random cytologic atypia is the sine qua non for diagnosis of that nevus. No other neoplasm in any organ is diagnosed on the basis of random cytologic atypia. Furthermore, in our view, none of the photomicrographs said by Clark et al. to exhibit atypia of melanocytes, random or not, exhibit any atypia of nuclei.¹² Clark and co-workers introduced this term, as they did "aberrant differentiation," in order to validate their concept of tumor progression of nevi to melanoma, but the nests do not "bud from the epidermis . . . in a random fashion," the "differentiation" is not aberrant, and the nuclei of melanocytes are not atypical. The relentless use of the term "dysplasia" also served the purpose of linking the most common of all melanocytic nevi to melanoma, and, in the case of the DNS, in a way that was confusing to physicians and alarming to patients. Last, "normal" nests of melanocytes is a misconstruction, and nests of melanocytes in the epidermis of DN are situated mostly at bases of rete ridges, just as they are in all other types of melanocytic nevi.

Clark et al.: "Three other histologic features, seen in the dermis, are commonly associated with melanocytic dysplasia: 1) lamellar fibroplasia, 2) concentric eosinophilic fibroplasia, and 3) patchy lymphocytic infiltrates."

Critique: The three histopathologic features just enumerated and affiliated commonly with DN are entirely non-specific; they also are found in some Spitz's nevi and in some melanomas. In fact, the fibroplasia may be encountered whenever proliferations of melanocytes occur at the dermo-epidermal junction of an epidermis that shows distinct rete ridges. We do not regard concentric and lamellar fibroplasia as two unrelated phenomena, but different ways of looking at the same finding; when collagen bundles are arranged in lamellae positioned immediately beneath melanocytes disposed mostly as solitary units at the dermo-epidermal junction along undulations of rete ridges, the appearance is termed (incorrectly) "concentric," whereas when collagen bundles are arrayed in lamellae seated immediately beneath nests of melanocytes stationed at bases of rete ridges, the appearance is described as "lamellar." In short, collagen bundles in both "types" are organized in lamellae, so-called concentric and lamellar fibroplasia being expressions of fibroplasia that is continuous. The differences in appearance are related to planes of section.

Clark et al. ". . . this copious lymphocytic response [in primary melanoma] is in contrast to the sparse, patchy infiltrate in a DN."

Critique: Infiltrates of lymphocytes may be sparse (or absent) in primary melanoma and dense in DN. The most dramatic example of the latter phenomenon is halo nevus, a Clark's nevus in most instances and one that involutes as a consequence of the effects on melanocytes of dense infiltrates of lymphocytes. In brief, density of infiltrates of lymphocytes (described badly by Clark and co-authors as "brisk" and "non-brisk") is irrelevant to diagnosis of either melanoma or DN.

Clark et al.: "In summary, melanocytic nevi with dysplasia have five distinctive histologic features: 1) persistent lentiginous melanocytic hyperplasia; 2) atypical melanocytic hyperplasia (melanocytic nuclear atypia); 3) lamellar fibroplasia; 4) concentric eosinophilic fibroplasia; and 5) sparse, patchy lymphocytic infiltrates. This constellation of histologic findings, especially when present in two or more lesions from a given person, is diagnostic of the DN syndrome."

Critique: These five criteria for histopathologic diagnosis of DN were quoted often as the gold standard by authors of articles about the subject published during the 1980s and 1990s. Unfortunately, the five criteria set forth by Clark et al are applicable equally to some thin melanomas. Furthermore, all five are not necessarily found together in many DN, e.g., the fibroplasia, and one is not a feature of that nevus at all, i.e., atypical melanocytic hyperplasia (melanocytic nuclear atypia). In short, the criteria established by Clark et al. for diagnosis of DN are flawed seriously. Even were they not, it is illogical on the basis of histopathologic findings in a nevus (irrespective of the number of nevi, e.g. 2 or 200), to leap to a diagnosis of a syndrome, particularly one that is reputed to be associated with a markedly increased risk of cutaneous melanoma. Not only is that illogical, it is hazardous to patients.

Clark et al.: "A patient with the DN syndrome may have lesions with these five cardinal features and, at the same time, common acquired melanocytic nevi. Other lesions may show only lentiginous melanocytic hyperplasia with concentric eosinophilic hyperplasia, but no melanocytic nuclear atypia. It is our opinion that the DN syndrome should not be diagnosed without the demonstration of melanocytic nuclear atypia, but both Sagebiel and Ackerman consider the abnormal pattern of melanocytic growth and the other histologic features sufficient for diagnosis without melanocytic nuclear atypia."

Critique: Clark and colleagues consistently used nuclear atypia of melanocytes as the sine qua non for distinguishing DN from CAN. Other students of these nevi could not recognize any nuclear atypia in what was said to be DN. Lynch and co-workers, in 1983, stated that they could not identify dysplasia in some FAMMM models.¹⁵ In that same year, Ackerman and Sagebiel, in their respective lectures at the Consensus Conference convened at the NIH, averred that nuclear atypia of melanocytes was not a feature requisite for diagnosis of so-called DN. Ackerman would soon propose that what Clark et al. considered CAN were simply smaller DN, i.e., DN that Clark et al. did not consider to be "funny looking." Never, however, did Ackerman subscribe to the concept of DN as elaborated by Clark et al.

Clark et al.: "The computations of Kraemer et al. suggest that such a patient [with type A DNS, i.e., with DN and no family history of DN or melanoma] has a risk of melanoma development 26 times that of the general population and a lifetime melanoma risk of approximately 20 percent. A type D2 patient has dysplastic nevi and at least two relatives with melanoma. This subset of patients with dysplastic nevi has a risk of melanoma 395 times that of the general population and a cumulative lifetime risk of 100 percent!"

Critique: Clark et al. repeatedly cite the data of Kraemer et al.16 about increased risk for melanoma among subtypes of the DNS. According to our calculations, all subtypes of DNS combined (except D-2), however, had no increased risk of melanoma when compared with the population at large. The data of Kraemer et al. were based on a presumed prevalence of DN in the general population of 2%, and an estimate, founded on histopathologic findings, of cutaneous melanomas in continuity with DN of 50%. Not only is the prevalence of DN in Caucasians more like 90%, but less than 25% of all melanomas that develop in Caucasians are in continuity with a pre-existing DN.^17–20 Practically no melanomas in Africans and Asians develop in nevi of any kind; they arise de novo.

Clark et al. also cited the data of Kraemer et al. relevant to patients with familial melanoma and DNS, namely, that such patients have a risk of melanoma 395 times greater than persons in the general population. But Kraemer and co-workers provided no evidence that DN are responsible for that risk for melanoma, except that 94% of the melanomas in those patients were said to arise in pre-existing DN. That is more than four times greater than our findings, the reason presumably being that Kraemer et al. interpreted findings of melanoma in situ as those of dysplasia. No definite conclusions can be drawn from the data of Kraemer et al. because those researchers did not define dysplasia, DN, and DNS in a lucid way and because they studied so few patients with familial melanoma and the DNS. Furthermore, if risk for melanoma in those patients were to be assessed accurately, it was necessary to have compared risk of melanoma among patients with familial melanoma who did and did not have DNS. The risk for melanoma in persons with familial melanoma can be attributed solely to genetics of melanoma; there is no need to invoke presence or absence of DN.

Clark et al.: "The foregoing data concerning heritability and risk are based on the histologic diagnosis of DN by the parameters set forth in this article, with the term melanocytic dysplasia used as a synonym for melanocytic nuclear atypia. This criterion is required to diagnose patients consistently as affected by the DN syndrome and to assess formally the risk of melanoma in putatively different types of patients. In this paper we use the term melanocytic dysplasia to include both persistent lentiginous melanocytic hyperplasia and melanocytic nuclear atypia. However, the sine qua non of melanocytic dysplasia remains melanocytic nuclear atypia."

Critique: In this article, Clark et al. state that "melanocytic dysplasia" is synonymous with "melanocytic nuclear atypia" and is the sole criterion for diagnosis of the DNS. In addition, nuclear atypia/"dysplasia" of melanocytes is claimed to be requisite for predicting risk of melanoma in patients with DNS. Having established the idea that "the term melanocytic dysplasia [is] used as a synonym for melanocytic nuclear atypia," Clark et al. proceeded to change their own definition of melanocytic dysplasia two sentences later when they wrote that "we use the term melanocytic dysplasia to include both persistent lentiginous melanocytic hyperplasia and melanocytic nuclear atypia." To confuse matters further, in the very next sentence, they state: "However, the sine qua non of melanocytic dysplasia remains melanocytic nuclear atypia." This series of sentences is unfathomable to us.

Clark et al.: "When melanocytic dysplasia and its (usually) associated mesenchymal changes are observed, the diagnostic terminology used in pathology reports is as follows: (1) junctional nevus with melanocytic dysplasia of the type seen in the DN syndrome; (2) compound nevus with melanocytic dysplasia of the type seen in the DN syndrome; (3) dermal nevus with melanocytic dysplasia of the type seen in the DN syndrome; or (4) de novo melanocytic dysplasia."

Critique: Five years later, in 1989, Clark would state, in a colloquy with Ackerman published in Seminars in Dermatology, that he no longer used the term "dysplastic nevus" in pathology reports, but employed instead "nevus with melanocytic dysplasia of the type seen in the DNS."12 But Clark and co-workers have never given a precise, comprehensible definition of "melanocytic dysplasia." In fact, in 1979, in an article about classification of "melanocytic dysplasia," Clark, in conjunction with Ainsworth et al., stated that "melanocytic dysplasia" could be found in congenital, recurrent, and halo nevi, and even in the dermal component of CAN.²¹ The term, as used by Clark, has no comprehensible meaning, and the same is true for the terms DN and DNS.

Clark et al.: "The term de novo melanocytic dysplasia is used to describe lesions that have the required histologic criteria but lack recognizable normal nevic tissue."

Critique: The authors do not make clear what they mean by "normal nevic tissue," perhaps because normal nevic tissue is a fiction; all melanocytic nevi are abnormal. Clark et al. believe that dysplasia may arise de novo and in association with junctional nevi, but they never inform readers how the two conditions can be distinguished from one another histopathologically. In our judgment, what Clark et al. call melanocytic dysplasia is either a junctional nevus entirely or melanoma in situ. There is no specific condition diagnosable histopathologically as "melanocytic dysplasia," anymore than there is junctional activity, atypical melanocytic hyperplasia, or pagetoid melanocytic proliferation. It seems that for Clark et al., "normal nevic tissue" resides in the dermis only and, therefore, nests of melanocytes at the dermo-epidermal junction of a DN are not regarded as those of a junctional nevus but as "de novo melanocytic dysplasia."

Clark et al.: "Table 2 (our Fig. 20) shows that the majority (55 percent) of superficial spreading melanomas have histologic evidence of precursor lesions. Of the 84 cases in which precursor lesions were present, 59 (70 percent) showed melanocytic dysplasia with or without associated nevic tissue."

Critique: It is curious that in 1969, in Cancer Research, Clark et al. stated that 9.6% of melanomas arose in a pre-existing nevus, but in 1984, Clark et al. wrote that 55% of superficial spreading melanomas originated in a pre-existing nevus.²² Despite this gross disparity, Clark and co-workers made no effort to explain the differences. In our experience, melanomas in Asians and Africans hardly ever begin in pre-existing nevi, and only 20–25% of melanomas in Caucasians develop in nevi of any kind, most of those being Clark's nevi.

Clark et al.: "In delineating histologic criteria for the evident steps of tumor progression and then using those criteria as the basis for the diagnosis of various lesional classes, we have made a strenuous attempt to correlate histologic features with clinical behavior. When this is done, the histologic features identify the likely clinical behavior of a lesion."

Critique: Histopathologic features frequently fail to predict the clinical behavior of melanocytic neoplasms. The consummate example is Spitz's nevus that may be affiliated with striking nuclear atypia, mitotic figures, and even abnormal mitotic figures, yet behaves in an entirely benign way. In contrast, some melanomas have slight nuclear atypia and few mitotic figures, but metastasize nonetheless. In the so-called DN, Clark et al. claim to find "random cytologic atypia" and on that basis link the nevus to progression to melanoma. We see no nuclear atypia and contend that far fewer than 0.01% of DN eventuate in melanoma; the vast majority of them are nevi for life.²³ Persons who have those nevi should not be told that that finding alone indicates a syndrome that puts them at markedly increased risk for melanoma. Except for those exceptional persons who have hundreds of large Clark's nevi, it is likely that diagnosis by a histopathologist of Clark's nevus (B-K mole, large atypical mole, DN, nevus with architectural disorder) does not necessarily imply any increased risk for development of melanoma. Physicians should protect patients, not scare them or injure them.

Our conclusion, just stated, seems to be verified by Clark himself when he wrote in 1995: "It would seem quite unwise to us to continue to use dysplastic nevus syndrome for the following reasons. The first, the term implies a common pathway for cutaneous malignant melanoma and dysplastic nevi. This may or may not be true. . . . Secondly, some formal definitions of syndrome state that multiple organ systems must be involved. Thirdly, familial melanoma may occur without dysplastic nevi, and in this context there can be no relationship between the two lesions."