Spitzoid melanoma in children: clinicopathological study and application of immunohistochemistry as an adjunct diagnostic tool

Authors


Eduardo Fonseca, MD, PhD, Servicio de Dermatología, CHU Juan Canalejo, Xubias de Arriba, 84, 15006 La Coruña, Spain
Tel: +34 981 17 80 00 ext. 296323
Fax: + 34 981 20 53 75
e-mail: fonseca@canalejo.org
fonseca@especialistasdermatologia.com

Abstract

Introduction:  The term spitzoid melanoma (SM) is reserved for a rare group of tumors with striking resemblance to Spitz nevus, often developing in children diagnosed in retrospect after the development of metastases.

Objectives:  To determine the biological significance of SM and to analyze the effectiveness of adjuvant diagnostic techniques.

Materials and methods:  A retrospective, observational study of 38 cases of SM in patients younger than 18 years. Histological type, Clark level and Breslow thickness, radial and vertical growth phase, mitotic count/mm2, ulceration, regression, vascular and perineural invasion, satellitosis, cytology and associated nevi were reviewed. An immunohistochemical analysis with HMB45 and Ki67 was performed in 10 cases. These features were correlated to patient’s stage and outcome.

Results:  Analysis of histological and immunohistochemical features should allow accurate diagnosis in most cases. Given the low mortality rate, no conclusions about the prognostic significance of histological parameters of the primary tumor could be established.

Conclusion:  We report the largest series of SM from a unique center. Although these patients may have a better prognosis than adults, some patients with SM develop metastasis and die, particularly after age 11 years. Therefore, we recommend using the same treatments as in adults.

Cutaneous melanoma in childhood is rare. Children and adolescents (age 0–17 years) accounted for only 1.3% of the cases of cutaneous melanoma in the USA during the past two decades.1 Only 0.3–0.4% of melanomas are diagnosed during the first decade of life,2 and even after including patients younger than 20 years, they only account for 1–4% of all melanomas.3

To render the ominous diagnosis of melanoma in a child is a very difficult decision for clinicians and pathologists. Relative lack of reliable pathological criteria for discrimination between benign and malignant melanocytic lesions combined with low frequency of cutaneous melanoma in childhood may lead to confusion or hesitance in diagnosis and thus cause delay in treatment.4–6

There is a group of melanomas cytologically characterized by large epithelioid and spindle cells, usually arranged in nests in the epidermis, with relatively low pagetoid upward migration. Because these are features typically seen in Spitz nevi, such lesions are sometimes referred as spitzoid melanomas (SMs). Furthermore, some of these lesions are initially diagnosed as Spitz nevi and are only recognized as melanomas when patients develop metastases.3,7

Several reports have been recently focused on clinical characteristics, pathological features and predisposing factors for development of cutaneous melanoma in childhood.3 However, because of the scarcity of these melanomas, there are only short series8–14 and isolated case reports15–17 of SMs in children. Studies of large series may be able to further determine the biological significance of these spitzoid lesions and analyze the possible effectiveness of adjuvant diagnostic techniques.

Materials and methods

Subjects and study design

A retrospective, observational study of all cases of SM in children and teenagers younger than 18 years old referred to UT – MD Anderson Cancer Center (MDACC) during the period 1992–2007.

Criteria for inclusion and evaluation

Although typical Spitz nevus is a benign lesion that can usually easily be distinguished from melanoma, there is no single criterion to establish a definite diagnosis of Spitz nevus or melanoma.3,9,18 As in some cases it seems impossible to render an unequivocal distinction, some authors10,19,20 have suggested that these lesions span a broad histological continuum, extending from benign to malignant tumors, where the risk of malignancy is proportional to the degree of deviation from the conventional features of Spitz nevus.

In order to minimize interobserver variability in the diagnosis, in this study, we used the following criteria and definitions for inclusion and evaluation of SM:

  • 1Clinical parameters: patients younger than 18 years.
  • 2Pathological parameters (examined on available slides):
  •  a. We defined histologically SM as a proliferation that retains some cytological attributes of Spitz nevus (‘spitzoid’ or ‘Spitz-like’ lesion) with a preponderance of atypical features belonging to melanoma:21
  • 1Spitzoid lesion: composed by spindle and/or epithelioid melanocytic cells, with these main features:18,22
  •    a. Epithelioid cells: round, large, occasionally multinucleated cells, with ground-glassy, angulated, eosinophilic cytoplasm and intranuclear pseudoinclusions, arranged in clusters or nests.
  •    b. Spindle cells: fusiform cells arranged in fascicles, with single, vesicular nucleus and eosinophilic nucleolus.
  • 2Atypical features: size >10 mm, asymmetry, lateral borders poorly demarcated (i.e. lack of circumscription), intra-epidermal pagetoid spread of single cells prominent in the center or else present at the periphery of the lesion, irregular intra-epidermal nest formation, ‘consumption’/atrophy or ulceration of the epidermis, deep extension, expansive rather than infiltrative deep border, absence of maturation (defined as lack of progressive decreasing size of nests and cells toward the deeper dermis), diffuse, non-nested sheets of cohesive cells in the dermis (without collagen bundles between cells), cellular atypia (pleomorphism, nuclear hyperchromatism) and deeply located or atypical mitoses. All these features were assessed as suggestive of malignancy and globally evaluated; none of them were independently considered a cause for inclusion or exclusion as melanoma of a spitzoid lesion.
  •  b. Prognostic indicators of SM were defined using the same criteria as for melanomas in adults:
  • 1Thickness was evaluated according to the method of Breslow,23 measuring from the most superficial aspect of the granular layer (or from the base of the ulcer, if it was an ulcerated melanoma) to the deepest point of invasion of the tumor.
  • 2Radial growth phase (RGP) and vertical growth phase (VGP) was determined by the histological parameters defined by Clark et al.24,25
  •    a. RGP: melanoma cells are present in the epidermis three rete ridges beyond the dermal component.
  •    b. VGP: aggregates of cells are seen in the dermis, forming nests larger than any intra-epidermal nest or there are mitotic figures in the dermal component.
  • 3Tumor infiltrating lymphocytes were classified in three categories:24,25 (i) ‘brisk’: lymphocytes invading throughout the whole VGP or extending across its entire base, (ii) ‘non-brisk’: lymphocytes focally infiltrating VGP, (iii) ‘minimal’ if there were only few lymphocytes. We used ‘minimal’ instead of the original category ‘absent’ because at least a small amount of lymphocytes was present in all specimens.
  • 4Ulceration was defined as the absence of epidermis overlying a portion of the primary melanoma based on microscopic examination of the histological sections.25 When present, the width of the ulceration was also recorded.
  • 5Regression was characterized by absence of melanoma cells in a focal region of RGP adjacent to VGP, widening of papillary dermis with increased collagen fibers parallel to surface, neovascularization, lymphocytes infiltrating diffusely, sprinkling of melanophages in papillary dermis and effacement of rete ridge pattern as described by Elder.24–26
  • 3Immunohistochemical parameters: increased proliferative activity in the deep areas of the lesion, as assessed by anti-Ki-67 antibody; absence of ‘maturation’ as evidenced with HMB45 (full-thickness or patchy labeling instead of only labeling the junctional and the more superficial dermal component).27,28
  • 4Follow-up features: patients who developed a metastatic event, defined as follows:19
  •  a. Regional metastases:
  • 1Satellite or in-transit metastases: metastatic foci of tumor cells similar to those in the primary tumor, situated around the site of primary tumor or between the primary tumor the and the regional lymph nodes.29
  • 2Lymph node metastases: defined as metastatic foci of tumor cells similar to those in the primary tumor and located into subcapsular sinus or parenchymal area of the node. Benign melanocytes with a nested pattern and primarily confined to the node capsule were defined as nodal nevus.

The lymph node metastases were classified as microscopic (clinically occult and detected pathologically by sentinel or elective lymphadenectomy) or macroscopic (clinically apparent by physical or radiological examination), both verified histologically.

  •  b. Widespread, distant metastases:
  • 1Subcutaneous tissue and distant skin/lymph nodes (clinically apparent by physical or radiological examination and verified pathologically): defined as metastatic foci of pleomorphic tumor cells similar to those in the primary tumor.
  • 2Visceral metastases verified by imaging and/or pathological studies.

Included in this study were patients younger than 18 years with lesions considered SMs according to a global evaluation of the above expressed criteria or showed metastasis. Patients with lesions sharing some classic features of Spitz nevus and some of the abnormal pathological features described above, in which a definite diagnosis of nevus or melanoma was not established and in whom a metastatic event was not developed during the follow-up period, were termed atypical Spitz tumors10 (with unpredictable biological potential) and were rejected from our study.

Protocol of treatment

Patients with melanoma thickness less than 1.0 mm, Clark II–III, non ulcerated, without regression or VGP underwent local excision with 1 cm of lateral margin. In addition to the wide local excision (WLE), cases with Breslow thickness less than 1.0 mm but Clark IV–V or presence of ulceration/regression/VGP also underwent a sentinel lymph node dissection (SLND). Melanoma cases with thickness greater than 1.0 mm underwent local excision with 2 cm of lateral margin and SLND. Patients with positive SLND or macroscopic lymphadenopathy (confirmed by fine needle biopsy) were treated with elective lymph node dissection (ELND). Patients with lymph node or visceral metastases received chemotherapy mainly with interferon-α (IFN-α), but in high-risk cases (>3 lymph nodes, extranodal extension or >3 cm lymph node), dacarbazine, cyclophosphamide and cisplatin were also used.

From 38 patients diagnosed of SM at UT – MDACC, 23 were treated in this institution. The remaining 15 patients were not entirely treated at our center, so we cannot be certain whether they received treatment consist with our protocol.

Data collected

Data were entered using a Microsoft Excel© database. Clinical details, follow-up data and outcome were obtained from the hospital files. For outside patients, UT – MDACC provided second opinion evaluation of tissue samples submitted by other physicians and pathologists. In those cases, we obtained follow-up information by email or phone inquiry to the referring institutions. The demographic data included were name, gender, race, age at diagnosis and family history. Special attention was paid to clinical appearance, location of the lesion and possible risk factors (sunburn, congenital and dysplastic melanocytic nevi, immunosuppression and other malignancies).

Hematoxylin-eosin sections of the primary lesions available from our files and requested from the referring physicians were evaluated by two of the authors (V. G. P. and S. P.) unaware of the clinical status of the patients. The following prognostic parameters were reviewed: histological type, Clark level and Breslow thickness, RGP and VGP, mitotic count/mm2 in the dermal component, ulceration, regression, blood vessel and perineural invasion, microscopic satellitosis, predominant cytology, associated nevi and surgical margins. Also reviewed were the available re-excision material as well as possible subsequent sentinel lymph nodes biopsies (SLNB) and regional lymphadenectomy specimens.

To improve the diagnostic accuracy and to study the immunohistochemical profiles of childhood SM (especially in borderline and difficult cases), an immunohistochemical analysis was performed on formalin-fixed, paraffin-embedded sections of available blocks or unstained slides in 10 cases. An antibody to S-100 protein (Dako, Carpinteria, CA, USA) was applied to visualize growth patterns and total tumor extent in 10 cases; the pattern of maturation was assessed in nine cases with an antibody against gp100 (with HMB45). Tumor cell growth fraction was measured in nine cases as numbers of Ki-67-positive cells/mm2 of tissue following a method previously reported.30 Anti-p16 and anti-CD99 were used in four and three cases, respectively.

The staging of patients was established according to the American Joint Committee on Cancer (AJCC) staging system for cutaneous melanoma 2001,15 based upon the tumor thickness and ulceration (T category), number of lymph nodes involved and tumor burden in nodes (N category), site of distant metastases and level of serum lactate dehydrogenase.

Also reviewed were the dates of initial surgery, WLE, SLND, elective lymphadenectomy and adjuvant chemo or radiotherapy. Duration of follow-up, disease-free survival and global survival were defined in months from the time of diagnosis to the latest uneventful contact, relapse or death, respectively.

Management of data and statistical analyses

A descriptive study was conducted for all the variables included in the study. Quantitative variables were expressed as average ± standard deviation and qualitative ones as absolute value, percentage and 95% confidence interval.

Prognostic parameters and immunohistochemical features were correlated to local recurrences, data of appearance, site of lymph node and distant metastases, patient’s stage and outcome. Comparison of averages was made using the Student’s t-test or Wilcoxon-Mann-Whitney test. Chi-square Statistic was applied for analysis of the association between qualitative variables. The significance level established in all the analysis was p < 0.05. A survival analysis was performed using the Kaplan-Meier test.

Ethics and policy

The study was approved by the Institutional Review Board at MDACC.

Results

From 1992 to 2007, 38 cases of SM were obtained from 21 (51.3%) male and 17 (49.7%) female patients. Table 1 presents the most outstanding clinical characteristics of our cohort. Ages at the time of diagnosis ranged from 1 to 17 years (average age 9.9 years). Ethnicity was known in 31 cases, with Caucasian the most frequent (87.1%). The anatomic site of the primary melanoma was available in 36 (94.7%) cases. The extremities were the most common site (16; 44.5%). There was no significant association between sex and age or sex and location.

Table 1.  Clinical characteristics of 38 patients with spitzoid melanoma
Variable
 nAverage (SD)Median
Age (years)389.9 (4.6)12.0
 n%95% CI
Ethnic group
 Caucasian2787.169.2–95.8
 Black26.51.1–22.8
 Hispanic26.51.1–22.8
Sex
 Male2155.338.5–71.5
 Female1744.728.9–61.5
Location
 Head and neck822.210.7–39.6
 Upper extremities616.77.0–33.5
 Trunk1233.319.1–51.0
 Lower extremities1027.814.8–45.4
Associated risk factors
 Congenital nevus0
 Phototype I–II17.10.4–35.8
 Immunodeficiency17.10.7–35.8
 Dysplastic nevus535.713.9–65.4
 Dysplastic nevus syndrome214.32.5–43.8
 Familial melanoma17.10.4–35.8
 Multiple melanoma214.32.5–43.8
 Multiple nevi642.918.8–70.3
 Other malignancies17.10.4–35.8

Relevant information about the clinical diagnosis at the time of the initial biopsy was limited because some patients were referred after a biopsy had been taken at a different institution. In some children, melanomas were clinically diagnosed as nevi. Pigmented lesions were removed because of increases in size, changes in color or both. Only one patient had history of a dysplastic nevus at the site of the subsequent melanoma, which was confirmed in the subsequent excision.

Risk factors were gathered in 24 patients; 5 of them had dysplastic nevus syndrome, 1 patient had a family history of melanoma and 2 had familial dysplastic nevus syndrome. In six patients (42.9%), multiple nevi were detected on physical exam, and one child had I–II Fitzpatrick phototype. None of the children had giant congenital melanocytic nevus, evidence of transplacental spread or a congenital malignant melanoma. One patient had been previously diagnosed with a brain stem glioma and another with juvenile rheumatoid arthritis and ulcerative colitis. None had a previous history of severe sunburn prior to the development of melanoma, immunosuppressive treatment or chemotherapy. The patient with the history of a brain stem glioma received local radiotherapy treatment.

Two patients had multiple melanomas. A 14-year-old girl patient presented with two primary melanomas at the time of diagnosis: an invasive SM on the leg and an in situ, non-SM on the arm. A 9-year-old boy (previously diagnosed with dysplastic nevus syndrome and familial dysplastic nevus syndrome) developed an invasive SM, and subsequent surveillance revealed two invasive melanomas on the arm and back and two melanomas in situ on the heel and forearm; none of the two were SM.

Data for histological classification of melanoma were available in 35 cases (Table 2). All cases were invasive. Nodular melanoma was the most common variant encountered in 17 (44.7%) cases, followed by superficial spreading melanoma in 8 (21.1%) and desmoplastic melanoma in 1 (2.85%) case. Ten (27.8%) lesions were considered unclassifiable because there was not enough epidermis beyond the dermal component available to examination. In two cases, the outside pathology diagnosis was Spitz nevus (both were associated with subsequent lymph node metastases). There was no significant relation between histological type and location.

Table 2.  Histopathological features of 38 spitzoid melanomas
Variable
Qualitative variablesn%95% CI
 Histological type
  Superficial spreading melanoma822.811.0–40.5
  Nodular melanoma1748.631.7–65.7
  Unclassified1028.615.2–46.5
 Clark level
  I0
  II38.32.2–23.6
  III822.210.7–39.6
  IV2261.143.5–76.4
  V38.32.2–23.6
 Radial growth phase827.613.4–47.5
 Vertical growth phase3090.974.5–97.6
 Ulceration825.012.1–43.7
 Lymphocytic infiltrate
  Minimal927.313.9–45.8
  Non-brisk2163.645.1–79.0
  Brisk39.12.4–25.5
 Regression26.11.1–21.6
 Vascular invasion515.25.7–32.7
 Perineural invasion39.12.4–25.5
 Associated nevus13.00.2–17.5
 Satellitosis0
 Margins
  Negative (tumor at ≥ 1 mm of peripheral margin)822.210.7–39.6
  Tumor present at ≤ 1 mm of peripheral margin513.95.2–30.3
  Positive2363.946.2–78.7
 Intensity of HMB45
  Mild444.415.3–77.3
  Moderate333.39.0–69.1
  High222.23.9–59.8
 Maturation with HMB45222.23.9–59.8
Quantitative variablesnAverage (SD)Median
 HMB45 %1018.1 (16.4)12.5
 Breslow thickness363.0 (1.8)2.8
 Mitotic figures/mm2323.5 (3.3)2.0
 Ki67 cells/mm2925.6 (19.3)17.0

Clark level and Breslow thickness were measured in 36 cases. There were 3 (8.3%) Clark level II cases, 8 (22.2%) level III, 22 (61.1%) level IV and 3 (8.3%) level V cases. Breslow thickness ranged from 0.40 to 8.0 mm (average 3.0 mm). These figures may underestimate the true Breslow thickness because the initial lesions were incompletely removed in 10 cases. Although there was no significant association of thickness with location, lesions located on the trunk were thinner than those at other sites (Fig. 1).

Figure 1.

Location of spitzoid melanoma and Breslow thickness.

Data on RGP and VGP were available in 33 cases. Thirty lesions (90.9%) had VGP. Mitotic figures were noted in 32 cases (Fig. 2), ranging from 1 to 13/mm2 (average 3.5); in four cases, mitotic figures were not identified in the examined sections, so the count was reported as ‘less than 1/mm2’. In one case, the mitotic count could not be measured because of the small dermal component, which extended to less than 1 mm. Eight (25%) lesions showed ulceration; as expected, a high number of them (5) were associated with lymph node metastasis. Regression was identified in two (6.1%) cases one of them with positive SLND. Lymphovascular and perineural invasion were present in five (15.2%) and three (9.1%) of 33 cases, respectively; four of them had further positive SLND (Figs. 3 and 4). Microscopic satellitosis was not detected in any of the cases. Lymphocytic infiltrate was analyzed in 33 cases. Brisk lymphocytic response was seen in 3 (27.3%) cases, non-brisk in 21 (63.6%) cases and minimal in 9 (27.3%) cases. Of the WLE specimens, residual melanoma was present in 12 (31.6%) cases.

Figure 2.

Spitzoid lesion arose on the head of a 17-year-old female. She was originally diagnosed with Spitz nevus and 3 years later developed parotid metastases. A) Low power (H&E, ×2): size < 10 mm, symmetry, lateral borders poorly demarcated, irregular intra-epidermal nest formation, ‘consumption’/atrophy of the epidermis. B) High power (H&E, ×20): expansive rather than infiltrative deep border, cohesive cells in the dermis, cellular pleomorphism and deeply located mitoses. H&E, hematoxylin and eosin.

Figure 3.

Lesion located on the arm of a 9-year-old male. A) Low power (H&E, ×2): highly pigmented nodular melanoma, Breslow 3.60 mm, Clark V, vertical growth phase. B) Vascular invasion (H&E, ×20). C) High power (H&E, ×40): deep border with cellular pleomorphism and mitoses. H&E, hematoxylin and eosin.

Figure 4.

Sentinel lymph node biopsy of the same patient in Fig. 3, located at ipsilateral axilla. A) Low power (H&E, ×4): metastatic foci of 5 × 3 mm in subcapsular and intraparenchymal area. B) High power (H&E, ×40): cells are similar to those in the primary tumor. H&E, hematoxylin and eosin.

Immunohistochemical studies were performed on 10 cases. Intense and diffuse HMB45 expression was seen in two of nine (22.2%) cases and mild or moderate in the remaining seven (77.7%). The percentage of HMB45-positive cells ranged from 1% to 50% (average 18.1%). Only two of these nine cases (22.2%) showed maturation (loss of labeling with decrease in the dermis) (Figs. 5 and 6). The KI67 labeling index of the invasive component was measured in 32 specimens and ranged from 4 to 65/mm2 (average 25.6/mm2) (Figs. 5 and 6). In both cases in which HMB45 seemed to show maturation, there was increased Ki67 expression, thus supporting a diagnosis of melanoma. No correlation with Breslow thickness, stage or survival was detected in statistical analysis.

Figure 5.

Immunohistochemical analysis. HMB45 and Ki67. A) Spitzoid melanoma (SM) showing patchy labeling with HMB45. Notice not only the strong intra-epidermal component (top) but also the positive cells in the dermis (bottom) (HMB45, aminoethyl carbazol and hematoxylin as counterstain). B) SM with strong labeling by HMB45 at the deep (transected) border of the lesion (deep reticular dermis) (HMB45, aminoethyl carbazol and hematoxylin as counterstain). C) Same lesion as in 5B. Numerous cells expressing Ki67 in the deep dermal component (MIB1, diaminobenzidine and hematoxylin as counterstain).

Figure 6.

Immunohistochemical analysis. HMB45 and Ki67. A) Low-power image of HMB45. Notice the strong labeling of the intra-epidermal component (up), while only a few cells in the dermis (down) express this marker (HMB45 and aminoethyl carbazol, ×10). B) Same lesion as in 6A. Intermediate power view of MIB1 labeling in the deep region of a spitzoid melanoma. Notice the large number (more than 10% of cells) of the melanoma cells expressing this marker [MIB1 (anti-Ki67) and aminoethyl carbazol, ×20].

Of the 38 patients in the study, SLND was performed in 25 (71.4%) cases. The procedure was not performed in nine cases, in four of them because of the presence of clinically positive lymphadenopathy at the time of diagnosis. In four cases, the information was not available. Fourteen SLND (56%) were positive (Table 3). The most frequent location of the metastatic foci was the subcapsular area (53.8%) (Fig. 4). There was no significant correlation between Breslow thickness and positive sentinel lymph node (SLN). Fifteen (51.7%) patients underwent ELND, eight of them had had previous positive SLNB and four had clinically evident lymphadenopathy. Thirteen (50%) patients received chemotherapy with IFN-α, which was interrupted in three cases because of side effects (hypertriglyceridemia, learning problems and fatigue). One of them received a combination chemotherapy regimen with IFN, etoposide, cyclophosphamide, vincristine and actinomycin.

Table 3.  Staging and follow up of 38 spitzoid melanomas
Variable
  1. ELND, elective lymph node dissection (lymphadenectomy); SLND, sentinel lymph node dissection; WLE, wide local excision.

Qualitative variablesn%95% CI
 SLND2571.453.5–84.8
 Positive SLND1456.035.3–75.0
 Location of metastatic foci in SLND
  Subcapsular753.826.1–79.6
  Intraparenchymal17.70.4–37.9
  Both (subcapsular and intraparenchymal)323.16.2–54.0
  Extracapsular17.70.4–37.0
  Not defined17.70.4–37.0
 ELND1551.732.9-70.1
 Positive ELND746.722.3-72.6
 Chemotherapy1350.030.4-69.6
 Stage
  IA516.16.1–34.5
  IB26.51.1–22.8
  IIA26.51.1–22.8
  IIB516.16.1–34.5
  IIIA825.812.5–44.9
  IIIB825.812.5–44.9
  IV13.20.2–18.5
 Final status
  Alive without disease3191.275.2–97.7
  Alive with disease12.90.15–17.1
  Dead25.91.0–21.1
  Missing4 
Quantitative variablesnAverage (SD)Median
 Time (days) between initial surgery and SLND2545.0 (39.2)33.0
 Time (days) between initial surgery and WLE2735.1 (19.3)33.0
 Time (days) between SLND and ELND1012.3 (9.0)14.0
 Time of follow up (months)3137.9 (42.1)28.2

Staging was established in 31 patients (Table 3). At the time of diagnosis and lymphadenectomy, more than 50% of patients were at stage III.

Follow up ranged from 0.46 to 176.9 months (average 37.9). Thirty-one patients were alive without disease, and four patients were lost in follow up. A survival analysis performed with Kaplan-Meier methodology is shown in Fig. 7.

Figure 7.

Survival probability of children with spitzoid melanoma.

Among the patients with events after SLNB, a 14-year-old male was alive with disease. He had a relapse 117 months after the initial diagnosis, with metastases in soft tissue, brain, liver, spleen, lung and humerus. He is currently receiving biochemotherapy and radiotherapy. Two patients in stage IIA deceased. One of them was a 15-year-old male with a previous brain stem glioma. He was treated with surgery and radiotherapy, with complete resolution. Two years and five months later, he had an amelanotic melanoma around radiated skin. Through a period of 28 months, he developed lung bilateral metastases of his melanoma, and 10 months later, he died of septic shock during chemotherapy treatment. The other death occurred in an 11-year-old boy who had been originally diagnosed with a Spitz nevus. Three years later, he developed lymph nodes and lung metastases.

All cases with subsequent metastases had been diagnosed as melanoma in the review of the original lesion.

Discussion

Although Spitz nevus is considered a benign melanocytic tumor,18 the differential diagnosis with melanoma can sometimes be difficult. The term SM is reserved for a rare group of tumors with striking resemblance to Spitz nevus, often developing in young individuals that are occasionally diagnosed as melanoma in retrospect after the development of metastases.31

Given the profound difficulty of distinguishing these lesions from benign Spitz tumors, misdiagnosis is common and occurs in up to 40% of children eventually found to have the disease.32 A prudent approach is to ensure complete surgical removal of the original lesion and to monitor the patient for recurrence or metastases. Consequently, some studies have only included cases in children with metastatic spread because it is considered the only definite proof of malignancy.33

Nonetheless, dermatopathologists faced with melanocytic lesions in children have to decide whether they are malignant or benign.33 Main complicating matters are lack of uniformity of opinion to which features enable this discrimination and lack of studies for validation of presumptive criteria.9 In this study, we have included the most commonly used histological criteria.

Regarding clinical diagnosis, SM often have abnormal morphological features like size >5 mm, asymmetry and irregular pigmentation. However, there are no uniformly reliable distinctive morphological attributes.31 Lesions that appear at young age and are located on the thighs are more likely to be Spitz nevi than common melanomas;34 however, such clinical features are not helpful in the majority of cases.

Epidemiology

The clinical characteristics of patients with SM (Table 1) are in consonance with previous studies (Table 4). As in adults, childhood melanomas affect mainly the Caucasian population and have equal sex distribution.3,5,31 As in other series,8,9,12,35 the most common primary tumor sites were the extremities followed by the trunk. Pol-Rodriquez et al.36 reviewed the literature between 1949 and 2006 and identified 82 cases of childhood SM with regional and/or widespread metastasis. These children had the same average age at the time of diagnosis (9.9 years old) as our series.

Table 4.  Previous series of metastatic spitzoid melanomas*
AuthorYearnSex (M:F)Age at diagnoses (mean, range)Main locationMortality (%)
  • H&N, head and neck; NAV, not available information.

  • *

    We considered only the patients with metastases included in the reported series because no final diagnosis was rendered in some studies on ‘atypical Spitz tumors’.

Barnhill9199919NAI20.5, 2–47Leg (44%)37
Spatz et al.101999111.2 : 1NAINAINAI
Lohmann et al.72200251.5 : 118.8, 7–28H&N (60%)0
Su et al.21200380.6 : 114.1, 5–29Limbs (50%)0
Gill et al.35200492 : 16.0, 2–10Limbs (66.6%)NAI
Jafarian et al.14200541 : 39.25, 4–16Leg (50%)75
Ferrara et al.8200670.4 : 122.1, 2–46Limbs (73%)NAI
Lee et al.11200651.5 : 125, 14–46NAI40
Murali et al.77200862 : 415.1, 6–38Limbs (50%)0
Current series2008171.4 : 19.52, 1–17Limbs (53%)11.76

Risk factors

Many of known risk factors for developing melanoma may be also identified during childhood,8 such as family history of melanoma, large congenital nevi, numerous nevi (including dysplastic nevi) and immunosupression.37 In our series, such conditions were identified in 18 cases (47%), similar to previous studies.13 One of the patients had SM around radiated skin after receiving radiotherapy for a brain glioma 3 years before. We have found only one reported case of cutaneous melanoma in childhood related to radiotherapy in a child with rhabdomyosarcoma, who developed non-SM 2 years later at the border of the irradiated area.13

There is some controversy about the risk for progression of congenital melanocytic nevus to melanoma. Historical estimates of melanoma risk were exaggerated because of diagnosis of atypical melanocytic proliferations developing in congenital melanocytic nevus as melanoma.31 Although all sizes are susceptible to malignant transformation, conventional melanoma is most commonly observed in relatively small congenital melanocytic nevus, while a number of melanomas in giant lesions display small cells.31,38 None of our SM cases arose within a congenital melanocytic nevus, and we have not found previously reported cases in the literature.

Documented evidence of a precursor non-congenital nevus at the site of subsequent development of melanoma has been reported in 47% of cutaneous melanoma in children.3 Interestingly, we showed this association only in one (3%) case.

Tumor factors

The relevance of regression as a prognostic factor has caused considerable argument in the literature.39,40 As some authors41–46 defended that it correlates with an impaired prognosis in thin tumors, SLNB was recommended for thin melanomas if there was evidence of extensive regression (involving >50% of the lesion).47 VGP, ulceration, tumor infiltrating lymphocytes and high dermal mitotic rate have been associated with metastatic behavior in adults.10,24,37,46,48–51

However, in previous series of melanoma in childhood, the only significant parameter associated with development of metastases and survival was tumor thickness.13 Moreover, some authors13,31 concluded that it is the main prognostic parameter, so early diagnosis may also be crucial in this population. However, we could not find a significant association between any of these histological parameters and stage, outcome or survival.

Lee et al.11 reviewed 33 SM and found a similar average Breslow thickness (2.58 mm) to ours (3.0 mm). Pol-Rodriquez et al.36 found an average tumor thickness even higher (4.42 mm), probably because they only considered lesions that had already originated metastases.

Immunohistochemistry

Although the diagnosis of SM mainly rests on microscopic morphological features, immunohistochemistry has been used in an attempt to facilitate the diagnosis.31 Even if some Spitz nevi have a heterogeneous staining pattern,52,53 HMB45 usually shows a stratified pattern with diminished expression toward the base of the lesion (maturation) in nevoid melanomas with pseudomaturation54 and in Spitz nevi27 in contrast with the diffuse/patchy pattern of expression throughout SM.31 Consistent with this observation, in our series, we only found maturation with HMB45 in two (22.2%) SM. However, in both of them, there was increased proliferative activity with Ki67, supporting a diagnosis of melanoma.

Spitz nevi also have low proliferation rates (1–2%) when assessed with Ki-67 compared with much higher rates in melanomas (15–30%).19,28,55–58 Moreover, melanoma has nuclear labeling throughout the lesion, while Spitz nevi have more staining at the top of the lesion than at the bottom.27 In this study, we decided to measure the most proliferative areas and quantified the proliferation index as number of positive nuclei/mm2; the average in this series was 25.6/mm2, much higher than that seen in nevi.30 In addition, p27, S100A protein and Fas (CD95) expression58–60 were shown to be higher in Spitz nevi than in conventional nevi or melanoma.

However, an overlap in the staining pattern of silver-staining nucleolar organizing region,61 proliferating cell nuclear antigen,62,63 melan-A/melanoma antigen recognized by T-cells-1,64 cyclin D1,55,58 p-Akt,65 c-kit,66 c-myc,67 c-fos,68 telomerase20,69,70 and Bcl-258 has been reported, so these techniques seem to be not very helpful in distinguishing Spitz nevi from SM.71 Although loss of expression of p-1658 and diffuse staining with CD9965 were reported in SM in comparison with nevi, we did not observed significant differences (data not showed).

Stage and survival

Because of the rarity of childhood melanomas and particularly of SM, little data concerning their prognosis are available in the literature. The metastatic potential of SM is considered relatively low by some authors.9 However, in our study, more than half of the patients (54.8%) had lymphatic or distant metastases. It is possible that being ours a referral center, a relatively high number of patients (5; 13.2%) seek treatment after a diagnosis of metastasis. The prognosis of SM in children, despite metastasis, has been suggested to be better than that observed in adults; however, this assertion remains controversial:72 While some SM with regional lymph node metastasis show no further progression,73 this is not true for every patient.36

Pol-Rodriquez et al.36 found a 88% 5-year survival rate in children with SM between 0 and 10 years of age compared with 49% in those between 11 and 17 years of age, so they concluded that prepuberal age (≤10 years) may be associated with longer survival. Younger children had thicker tumors (4.67 vs. 3.93 mm) and a greater proportion of regional metastases (86%), but older patients had a higher rate of widespread dissemination (41%). In our series, both patients who died of the disease were older than 10 years (11 and 15 years old); we were not able to determine if both patients had already reached puberty.

Comparing with other series of childhood melanoma regardless of histological type, localized disease is more frequent (57–88%)5,13 than in this study. Mortality had a wide range, oscillating from 0.1% to 0.2%13 (as in our series) to almost 40%.3,5

Treatment

Many authors recommend that SM should be treated as the rest of melanomas.74 A complete excisional biopsy is preferred because incomplete excision by shave or punch biopsy prevents the assessment of the lateral edges and deep components.59,75 Afterward, WLE, SLND and long-term clinical follow up are advisable.16,19,76

Given the low morbidity of the SLND, it may also serve as an adjunct procedure in the evaluation of diagnostically difficult spitzoid tumors by increasing the sensitivity of the diagnosis and providing potentially useful prognostic information.72,77 Although negative SLND is not a guarantee against future recurrences or tumor spread, positive SLND is usually a clue of malignancy.19,21,72,76–78 In a review of 21 spitzoid tumors,77 significant higher Breslow thickness in patients with positive SLN was detected. Nonetheless, we could not prove significant differences between SLN-positive and SLN-negative cases with regard to histological features. In our series, 56% of patients had positive SLND; but so far, no cases of death were detected in patients diagnosed with positive SLND. Both deceased children had macroscopic lymph node metastases at the time of the diagnosis.

Some authors have suggested that patients with positive SLND should be treated with adjuvant therapies such as interferon.78 Also, combinations of chemotherapeutic drugs may be effective in children.79 In our series, the number of treated patients with widespread disease is too small to achieve statistical significance.

Immunohistochemical staining patterns supplemented the morphological features with the detection of either lack of maturation with HMB45 or with the presence of increased proliferation with MIB1. These features were helpful to establish the diagnosis of SM in some lesions previously diagnosed as Spitz nevus, including one of the cases that subsequently developed distant metastasis.

In conclusion, we report herein the largest series of SM from a unique center. We obtained valuable information about epidemiological and histopathological features of this controversial and still poorly understood malignancy. Careful analysis of histological features as well as the additional information provided by immunohistochemistry should allow accurate diagnosis in most cases. Given the low mortality rate, our study did not allow to draw conclusions about the prognostic significance of clinical or histological parameters of the primary tumor. Although from our data it seems that these patients may have a better prognosis than adults showing the same type of lesions, still a number of patients with SM will develop metastases and die of their disease, particularly when melanoma is diagnosed after age 11 years. At any rate, until further studies determine more accurately the prognosis in these patients, it seems logical for them to be treated using the same therapeutic approaches as in adults.

Acknowledgements

These authors thank Dr Maria Niveiro and Ignacio Aranda, Hospital Universitario de Alicante, Spain; Dr Peter Bozner, Prefered Anatomic Pathology Services, Lafayette, LA, USA; Dr Enrique Lerma, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain; Dr Bruce Edwin King and John Legere, Lancaster, PA, USA; Dr John Hegstrom, Centura Health/Penrose Dep. Pathology St Francis Health Services; Dr Cindy Braden and Dr Brock M. Bordelon, University of Colorado Cancer Center; Dr Ruth Y. Peng, Waukesha Memorial Hospital Laboratory and Dr Thomas Korkos; Dr Parsy, Mark Wienhpal and Matters, CO, USA; Dr Edward Perez, Regional Medical Laboratory, TX, USA; Dr Félix Contreras Rubio, Hospital Universitario La Paz, Spain and Dr Jesús Cuevas Santos, Hospital Universitario de Guadalajara, Spain

Ancillary