The tumorigenic phase is established by the presence of dermal aggregates of melanoma cells larger than the junctional malignant nests (hematoxylin-eosin, original magnification ×200).
Lentiginous single-cell proliferation of atypical melanocytes overlying an area of fibroplasias and regression (hematoxylin-eosin, original magnification ×200).
Low-power view of a metastasizing thin melanoma showing extensive areas of regression and melanoderma (hematoxylin-eosin, original magnification ×50).
Aggregates of melanin-laden macrophages (nodular melanosis) are noted within an area of regression (hematoxylin-eosin, original magnification ×100).
Expansion of the papillary dermis in an area of extensive regression and fibrosis with scattered lymphocytes and melanophages (hematoxylin-eosin, original magnification ×50).
Guitart J, Lowe L, Piepkorn M, Prieto VG, Rabkin MS, Ronan SG, Shea CR, Tron VA, White W, Barnhill RL. Histological Characteristics of Metastasizing Thin MelanomasA Case-Control Study of 43 Cases. Arch Dermatol. 2002;138(5):603-608. doi:10.1001/archderm.138.5.603
Copyright 2002 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2002
To study clinical and histological features associated with metastasizing thin melanomas (MTMs).
Case-control study of clinicopathological features of patients with MTMs by a panel of 10 dermatopathologists.
Members of the North American Melanoma Pathology Study Group selected the cases from the melanoma databases at 8 academic institutions.
Forty-three patients with MTMs (<1 mm thick) and 42 control subjects without metastasis matched for age, sex, tumor site, and Breslow thickness.
Main Outcome Measures
Clinical (age, sex, site of lesion, stage at diagnosis, metastasis site, disease-free survival, and outcome) and histological (Breslow thickness, Clark level, growth phase, regression, and inflammatory response) features of patients with MTMs vs controls.
There was an overrepresentation of axial tumors among patients with MTMs. Extensive regression was present in 18 patients (42%) with MTM vs 2 matched control subjects (5%) (95% confidence interval, 21%-53%; P = .001). Other histological variables were not significantly different. Two patients had melanomas in situ with subsequent metastasis.
Thin melanomas with extensive regression represent a group at higher risk for the development of metastasis. Furthermore, the risk of metastasis cannot be dismissed in cases of melanoma in situ.
MICROSCOPIC assessment of Breslow thickness is the most accurate prognostic indicator in malignant melanoma.1 There is ample evidence that tumor thickness is directly proportional to the development of metastasis and to the overall patient survival.2 In general, melanomas with a Breslow thickness of less than 1 mm have a good prognosis, with a low incidence of metastasis. However, thin melanomas may develop metastasis.3- 12 We have reviewed our experience with metastasizing melanomas with a Breslow thickness of less than 1 mm. Our goal was to identify those clinical and histological factors that may predict a subset of thin melanomas more likely to metastasize.
Electronic records were searched from the melanoma databases and pathology departments at University of Michigan, Ann Arbor; University of Illinois, Chicago; Duke University, Durham, NC; Vancouver University, Vancouver, British Columbia; the Connecticut Tumor Registry, Hartford; The Johns Hopkins University, Baltimore, Md; Wake Forest University, Winston-Salem, NC; and Northwestern University, Chicago. Histopathological material from patients with primary melanoma with a Breslow thickness of less than 1 mm and a known history of metastasis were selected. All metastases were confirmed by histological examination. All biopsy specimens from patients were reviewed, and it was found that they had negative surgical margins, or if the margins were involved, a complete reexcision specimen had been performed. Each case was evaluated by the contributing pathologists before submission to the study. The slides selected for panel evaluation were the most representative sections of the lesion. The slides were reviewed independently by a panel of 10 dermatopathologists with experience in pigmented lesions. The control group included 42 patients with thin melanomas with no evidence of metastasis after a minimum follow-up of 6 years. The control patients were matched for age (within a 10-year range), sex, site (extremities, truncal, or head and neck), and Breslow thickness. The diagnosis was confirmed in all lesions. The following factors were assessed: thickness, Clark level, and growth phase (tumorigenic vs nontumorigenic). Tumorigenic growth phase was defined by the presence of dermal mitosis or dermal aggregates of tumor cells larger than the nests in the epidermal component (Figure 1).13 Presence, extent (<50% or ≥50%), and evolutionary stage of regression were also assessed. The percentage of regression was not related to the overall surface area of the lesion, but was based on the percentage of linear regression on the most representative slide of the lesion. Regression was defined as dermal fibroplasia with an absence of epidermal and dermal involvement by melanoma cells, but allowing for lentiginous single-cell proliferation of atypical melanocytes along the dermal epidermal junction (Figure 2). Host response was defined as the presence of tumor-infiltrating lymphocytes intermingled among tumor cells or a lymphoid infiltrate at the base of the lesion. Other notable histological findings (coexisting lesion and angiolymphatic spread) were recorded. If agreement by all panelists was not achieved, a simple majority was considered sufficient for tabulation purposes. Follow-up information, including patients' current status, stage at the time of diagnosis, disease-free survival, and sites of metastasis, was obtained from tumor registries and from patients' medical records. t Tests and χ2 assays were used for statistical analysis. P<.05 was considered significant.
The study group consisted of 43 patients (23 men and 20 women), with a mean age of 50.0 years (range, 27-79 years) (Table 1). The back and shoulder were the most common locations for lesions, affected in 21 (49%) patients. The second most common location was the lower extremity (10 patients), followed by the upper extremity and chest, occurring in 4 patients in each location. Two had lesions on the ears and 1 patient each had lesions on the scalp and abdomen. Staging performed at the time of diagnosis revealed 27 (63%) patients without clinical evidence of metastasis (stage I). Lymph node metastases (stage III) were detected in 12 (28%) patients. Four patients (9%) had distant metastasis (stage IV) at the time of presentation. Eventually, metastases were detected in every patient. The most common site for metastasis was the regional lymph nodes, affected in 23 patients. Nine patients had metastasis to more than 1 lymphatic region or had lymphatic metastasis associated with systemic metastases. Six patients had metastasis involving soft tissue or skin. Lung or pleura was the site of metastasis in 4 patients and liver in 3. Two patients had metastasis reported in the stomach and bone. In addition, other sites of metastasis included the brain (2 patients) and parotid gland (1 patient). One case with in-transit metastasis also occurred. The mean follow-up was 52 months. The time to detection of metastasis was not available in most cases. Five patients reported as having developed metastasis had unspecified sites. When the study was closed, 26 patients had died with melanoma, and 12 patients were alive with disease. Five patients were alive without clinical evidence of melanoma.
The matched control cases included 42 patients (23 men and 19 women), with a mean age at presentation of 48.5 years (range, 23-72 years) and a mean follow-up of 13 years (range, 6-27 years) without evidence of metastasis. The lesions were located on the torso in 25 patients, extremities in 14, and head or neck in 3.
The panel agreed on the diagnosis of melanoma in all the cases. The mean ± SD thickness for the metastasizing lesions was 0.59 ± 0.20 mm (range, 0 [melanoma in situ] to 0.95 mm). There was some discordance in the assessment of Clark level. Consensus, defined as agreement with no more than 1 dissension within the panel, was reached in 15 (35%) of the 43 patients . In 3 of the cases, the panelists' assessments were evenly divided and a simple majority could not be reached. Patients with metastasizing thin melanoma (MTM) were classified as follows: level I, 2 patients; I or II, 1 patient; II, 10 patients; III, 22 patients; III or IV, 2 patients; and IV, 6 patients. The distribution was not significantly different from that of the control cases (P = .81). Consensus for the assessment of growth phase was reached in 39 patients (91%). Ten cases (23%) were determined to be in the nontumorigenic (radial growth) phase and 33 (77%) in the tumorigenic (vertical growth) phase. Consensus regarding growth phase could not be reached in 5 patients (12%). Dermal mitotic figures were noted in 8 cases, while large dermal aggregates of tumor cells were seen in all the tumorigenic cases. Regression was identified in 21 patients (49%) vs 8 control cases (19%) (P = .001) (Figure 3, Figure 4, and Figure 5). Extensive regression involving 50% or more of the lesion surface was noted in 18 patients (42%) and less than 50% involvement in 3 (7%). Regression was not identified in 22 patients. In the control group, 2 patients (5%) had extensive regression. The occurrence of extensive regression was significantly different between patients and controls (95% confidence interval, 21%-53%; P = .001). A mononuclear cell inflammatory infiltrate was prominent in 36 cases. The infiltrate was predominantly at the base of the melanocytic lesion in 9 cases and was mostly intermingled with the melanocytes (tumor-infiltrating lymphocytes) in 13. A combined infiltrate at the base with tumor-infiltrating lymphocytes was noted in 14 cases. Five cases had no significant inflammatory infiltrate. A coexisting melanocytic nevus was identified in 7 cases, including 2 congenital nevi and 2 dysplastic nevi. In 4 cases, dermal nevomelanocytic nests were noted, but extensive regression prevented further assessment of the nevi.
The mean thickness for the control specimens was 0.56 mm (range, 0-0.94 mm) (P = .70). Clark level distribution was II in 16 cases; III, 20; III or IV, 3; and IV, 3. Thirty cases (71%) were determined to be in the tumorigenic phase and 12 (29%) in the nontumorigenic phase (P = .62). Regression was absent in 34 control cases (81%) and was present in 8 (19%), with 6 showing limited regression (<50%) and 2 showing extensive regression (≥50%). A host response was absent in 5 cases, a lymphoid infiltrate at the base was noted in 26, and tumor-infiltrating lymphocytes were seen in 25. The inflammatory host response of the metastasizing cases vs the control group was not significantly different (P = .52). Seven control cases had a coexisting lesion, including 6 dysplastic nevi and 1 congenital nevus. None of the MTM or control cases were ulcerated.
In recent decades, a steady increase in the incidence of thin melanomas has been reported.14,15 This disturbing trend may, however, reflect the detection and removal of melanomas at early stages, when there is less risk for metastasis. Regrettably, thin melanomas have metastatic potential, with an estimated risk of 3% to 8%, depending on the length of follow-up.7,9,16- 18 Despite numerous studies on the subject, the clinical and pathological risk factors for these MTMs are largely unclear. This problem is partly because of variations in the defining criteria for thin melanomas used in prior studies,7,19 with Breslow thickness varying from 0.76 mm to 1.69 mm. Furthermore, some investigations combined recurring lesions with MTMs, further confounding the results.20 The histological criteria associated with high metastatic risk in thin melanomas have not been clearly elucidated. There is some evidence that the metastatic potential of thin melanomas, compared with that of thick melanomas, is significantly more affected by factors other than thickness. The size of the lesion as measured by surface area, the presence of regressive changes, and location may be important variables in the prognosis of thin melanomas.3,21- 24 Thin melanomas show regression more frequently than do thick melanomas.3,25 The presence of regression has been reported in 7% to 61% of all thin melanomas and in 40% to 100% of MTMs.3,18,26- 28 The variability in the reported incidence of regression in thin melanomas is probably again because of the lack of uniformity in defining criteria and the minimal area of involvement required for inclusion. In our study, tumor regression was the most conspicuous histological finding, encountered in 21 MTM cases (49%) vs 8 (19%) of the control cases. Furthermore, extensive regression involving 50% or more of the lesion was seen mostly in the metastasizing group (P = .001). Slingluff and colleagues7,26 found a similar rate (40%) of severe regression in MTMs, compared with 17% of nonmetastasizing thin melanomas. Other studies10,29 have identified a higher rate of metastasis in thin melanomas with regression (8%) compared with cases without regression (5%). Moreover, the 10-year survival for patients with thin melanomas with regression was 79% vs more than 95% in patients without regression.17,30 All patients from the Sydney Melanoma Unit cohort27 and University of Illinois database28 with thin melanomas and lymph node metastasis at presentation had extensive regression involving more than 75% of the tumor area. The poorer prognosis of MTM with extensive regression has been corroborated by other reports.21,23 However, less extensive regression was also noted in 61% of thin melanomas that never metastasized or recurred.27 Other investigators could not demonstrate an effect of regression on disease-free interval or survival.11,25,31- 34
Severe regression has been reported more frequently in back lesions and in men.3,7,31 Our findings concur with the overrepresentation of lesions from the back. However, we could not confirm the male predominance among patients with MTM.
Sondergaard and Hou-Jensen30 found that metastasis was correlated not only with the extent of regression but also with the thickness of it. Patients with regression but without thick scar formation had a better prognosis. The implication is that the regressed tumor was preceded by an originally thicker tumor.
We identified 2 cases of apparent melanoma in situ that subsequently developed metastasis. One of the cases had extensive regression involving more than 50% of the surface area. This observation suggests a preceding invasive tumor with metastatic potential, before regression of the dermal component. However, regression was not identified in the other melanoma in situ with metastasis. This perplexing case is difficult to reconcile with the notion that theoretical nontumorigenic lesions have metastatic potential. Although a thorough examination did not reveal a second primary lesion, a primary occult visceral melanoma cannot be excluded. Nonetheless, our results show that it cannot be assumed that an apparent level I melanoma will not metastasize. Furthermore, 10 of the MTM cases studied were histologically classified by the panel as nontumorigenic. Although consensus in the panel regarding growth phase was reached in 91% of the cases, there were cases in which growth phase status was difficult to determine. The overall high reproducibility of growth phase assessment has been corroborated by other recent studies.35,36 However, the clinical relevance of such determination is less clear, as shown by the 10 "nontumorigenic," yet metastasizing, cases in our series. Perhaps, we should resist from providing such categorical predictions that, in some instances, may not be more than an educated guess, yet with profound prognostic implications to the patient.
Contrary to prior reports, we did not find a high incidence of MTM lesions from the head and neck area, but did find a high number of axial tumors, primarily from the back. A higher incidence of axial lesions among MTM cases has also been observed by other authors.3,7,19,26 Two other studies37,38 also showed that patients with thin melanomas involving extremities had a lower metastatic rate and better prognosis.
In some studies,39- 42 thin melanomas from the head and neck have shown a higher rate of recurrence and metastasis. The higher rate of recurrence reported in head and neck melanomas may be associated with narrower surgical margins in anatomical sites cosmetically compromised and with difficult surgical repair. Another possible explanation is that the skin from the head and neck also has a thinner reticular dermis, which is often damaged by actinic changes. Hence, a thin Breslow level could be associated with a deep anatomical level of invasion (Clark level). Because there were only 3 head and neck cases in our cohort, we could not draw any definite conclusions about this subset of patients.
There is no reliable marker for the metastatic potential of thin melanomas. Studies of proliferation markers (MIB-1)43 and p53 expression44 of thin melanomas found no association between the proportion of positive cells and risk of metastasis. Bjornhagen et al45 showed statistically significant differences between MTM and nonmetastasizing thin melanoma in regard to the nuclear correlation coefficient, indicating that thin melanomas with pleomorphic and, possibly, densely packed nuclei were more likely to recur. A significant difference of mean nuclear volume between MTMs and nonmetastasizing thin melanomas has also been reported.17 Angiogenesis of thin melanomas has been investigated, with conflicting results. Although Guffey et al46 found no predictive value in measuring vascularity in thin melanomas, Barnhill and Levy47 suggested that the onset of angiogenesis is related to inflammatory regression and progression to the vertical growth phase.
Our study confirms that thin melanomas with extensive spontaneous regression represent a group at higher risk for the development of metastasis. Such patients should be carefully followed up, especially when the primary lesion involves the torso. However, in about half of our patients, we were unable to identify regression or any other histological or clinical distinguishing features. This should encourage further research to identify reliable molecular or serological prognostic indicators for patients with thin melanomas.
Accepted for publication June 20, 2001.
We appreciate the comments and contributions of Lynn From, MD; David Elder, MD; and Alfred W. Rademaker, PhD.
Corresponding author: Joan Guitart, MD, Department of Dermatology, Northwestern University Medical School, 675 N St Clair, Suite 19-150, Chicago, IL 60611 (e-mail: email@example.com).