Spitz Tumors in Children: A Grading System for Risk Stratification

Objective  To describe a grading system for risk stratification of atypical Spitz tumors in children and adolescents. In some circumstances, unequivocal distinction between Spitz nevus and melanoma is practically impossible. It is likely that these lesions for which we lack specific diagnostic criteria represent a broad histological continuum extending from benign to malignant tumors. Therefore, we propose that Spitz tumors be categorized into low-, intermediate-, or high-risk categories based on the accumulation of abnormal features.

Design  Retrospective study.

Settings  Institutional practice.

Patients  We present 30 cases of atypical Spitz tumors in patients younger than 18 years evaluated for at least 3 years or in whom a metastatic event developed during this period.

Intervention  None.

Main Outcome Measure  The grading system was formulated after data collection.

Results  Among the parameters studied, only diagnosis at age greater than 10 years, diameter of the lesion greater than 10 mm, presence of ulceration, involvement of the subcutaneous fat (level V), and mitotic activity of at least 6/mm² carried a likelihood ratio greater than 1.50 and were therefore used for the grading system.

Conclusion  The application of an objective grading system, such as the one described herein for the first time, is the first step in providing useful information for the management of atypical Spitz tumors.

THE TYPICAL Spitz nevus is a benign tumor that can be distinguished from melanoma in most instances. However, as stated by Allen and Spitz¹ in 1953, in some circumstances, unequivocal distinction between Spitz nevus and melanoma is practically impossible. It is likely that these lesions for which we lack specific diagnostic criteria represent a broad histological continuum extending from benign to malignant tumors. Therefore, we propose that Spitz tumors be categorized into low-, intermediate-, or high-risk categories based on the accumulation of abnormal features. We describe a scoring system based on data from a series of 30 patients aged 17 years or younger who presented with an atypical Spitz tumor.

The 30 patients with a histopathologic diagnosis of atypical Spitz tumor whose data formed the basis of the scoring system were less than 18 years of age when the diagnosis of atypical Spitz tumor was made, were evaluated for at least 3 years or developed a metastatic event during this period, and had representative slides available for review. The cases were collected from 1959 through 1997, the range of earliest and latest reports. The slides came from the regular or consultation files of the following institutions: Institut Gustave-Roussy, Villejuif, France; St John's Institute of Dermatology at St Thomas' Hospital, London, England; Children's Hospital and Brigham and Women's Hospital, Boston, Mass; and the Johns Hopkins Medical Institutions, Baltimore, Md. Twenty-six cases had been previously reported, including 8 cases from Institut Gustave-Roussy,² 11 cases from St Thomas' Hospital,³ and 7 cases from Children's Hospital, Brigham and Women's Hospital, and the Johns Hopkins Medical Institutions.⁴ The term atypical Spitz tumor was used in this study to designate lesions sharing some but not all of the classic features of Spitz nevi or displaying unusual features.^4-6 These features are summarized in Table 1.

Clinical variables included age at diagnosis (≤10 vs >10 years), sex, lesion site (axial vs extremities), metastatic event, date of last follow-up, and vital status.

In order to provide reproducible criteria to categorize the lesions, only 5 quantitative histological or well-defined parameters were evaluated. The lesion diameter (≤10.0 vs >10.0 mm) was measured on the slide by micrometer. The tumor thickness (≤3.00 vs >3.00 mm) was measured by micrometer according to the method of Breslow.⁷ Ulceration was defined as an interruption of the whole epidermis, including the stratum corneum. The mitotic rate in the dermal component (0-5, 6-8, or >8/mm²) was recorded after the lesion had been scanned for hot spots.⁸ The anatomical level (I-IV vs V) was recorded according to the method of Clark et al.⁹

The scoring system was derived from a series of simplified likelihood ratios.¹⁰ The scoring system represents a series of rounded ratios between the conditional probabilities for whether each parameter state is associated with metastasis. When the ratio was lower than 1.5, the assigned value was 0.

The 30 patients with sufficient data available to be included in the study included 11 patients with histologically documented metastases and 19 patients without metastasis after a follow-up period of at least 3 years.

Table 2 summarizes the distribution of patients and the ratios of conditional probabilities for each parameter. Among the parameters studied, only age at diagnosis greater than 10 years, diameter of lesion greater than 10.0 mm, presence of ulceration, involvement of the subcutaneous fat (level V), and mitotic activity of at least 6/mm² carried a ratio of conditional probabilities greater than 1.50 and were therefore used for the grading system.

The grading system is presented in Table 3.

The lesions were then graded and categorized. Table 4 summarizes the distribution of cases by category of risk. Among 15 cases categorized as Spitz tumor with low risk of metastasis, only 1 case was characterized by lymph node metastasis and death. The patient was a 15-year-old boy with a lesion on the neck; he died 7 months after diagnosis.⁶ Histologically, the tumor measured 4.4 mm in diameter and 2.95 mm thick, involved the reticular dermis but not the subcutaneous fat, and displayed a mitotic rate of 5/mm². The morphologic appearance suggested an atypical Spitz nevus and a halo nevus (Figure 1). Three of 6 cases categorized as intermediate risk and 7 of 9 categorized as high risk (Figure 2) had metastases.

We describe a grading system for the categorization of Spitz tumors in children and adolescents. This scoring system is based on our experience with 30 cases with metastasis and/or follow-up of at least 3 years.

Spitz nevi are defined by a set of histological features that has evolved from the initial description by Sophie Spitz in 1948.¹¹ There is a category of lesions that do not demonstrate the classic features of Spitz nevi or that exhibit atypical features. Since no set of criteria can predict clinical outcome with certainty in this group of lesions, the dichotomous categorization into nevus and melanoma appears artificial. Moreover, the final diagnosis is often based on the degree of deviation of these lesions from the classic Spitz nevus. Therefore, it is not unexpected that the threshold to make a diagnosis of melanoma in the group of atypical Spitz tumors should vary from one pathologist to another and that these lesions constitute one of the most important sources of discordance among pathologists in the field of the melanocytic lesions. We propose the separation of Spitz tumors into 3 groups: unequivocal Spitz tumors without any abnormal or unusual features, melanomas, and atypical Spitz tumors. For the latter category we propose a grading system for the categorization of risk for metastasis as low, intermediate, or high.

There are 2 main justifications for the use of a grading system: to improve on the predictive value of conventional histological analysis and to increase the reproducibility between observers for diagnosis and assessment of prognosis. With respect to Spitz tumors and related lesions, 2 antithetical views are presented in the literature. According to one view, most of the lesions diagnosed as Spitz nevi in reality deviate from the generally accepted criteria and therefore represent a phenotypic continuum between the Spitz nevus and melanoma.¹² According to another perspective, the Spitz nevus is a completely benign lesion that has no relation to melanoma biologically and in many cases displays features that permit an unequivocal diagnosis of Spitz nevus.¹³ The reports of a fatal outcome from lesions not distinguishable from Spitz nevi support the first position.^6,14-18 It is our view that even though the latter cases are quite rare, their existence justifies the idea that all Spitz tumors have some potential risk for metastasis. Perhaps more frequently encountered are the group of Spitzlike lesions with metastatic evolution, for which no definitive diagnosis of nevus or melanoma could be made a priori. It is likely beyond the scope of routine histopathology to render a definitive diagnosis for these lesions. The problem is illustrated in the present study by the case with metastasis and death, which was diagnosed as a nevus according to the classic dichotomous system. This lesion would have been classified as an atypical Spitz tumor with low risk of metastasis according to our grading system. Several authors have recently suggested that the current histological criteria for diagnosing melanocytic neoplasms are highly prone to variable interpretation and are entirely subjective,^19,20 indicating a lack of reliability. These statements are even more applicable to the specific category of Spitz tumors. Consequently, we believe that our grading system should promote better reproducibility among observers for the histological evaluation of Spitz tumors.

The parameters evaluated to establish the grading system were either quantitative or easily definable. Seventy percent of Spitz tumors occur in patients younger than 20 years, and among these, half are diagnosed before age 10 years.²¹ On the other hand, melanomas are extremely rare in children less than 10 years of age. It is estimated that 94% of Spitz nevi are less than 1 cm in diameter.²¹ Most of the reported Spitz tumors with metastases measure more than 1 cm. In the present series, fat involvement was observed more frequently in patients with metastatic evolution. Since, as previously reported, tumor thickness is the major prognostic factor in childhood melanoma,² it is likely that fat involvement is a correlate of tumor thickness, which failed to have prognostic value because of the small series of patients. In most of the comparative studies attempting to identify the most discriminant features for the differential diagnosis of Spitz nevi and melanoma, ulceration was frequently associated with metastasis. Mitoses may be absent, rare, or numerous both in Spitz tumors without metastatic evolution and in fatal melanomas. However, a high dermal mitotic rate was often associated with metastatic evolution. Deeply located mitoses have also been ascribed to melanoma. However, the evaluation of depth of a mitosis is often subjective and related to tumor thickness.

Since the cases in this series were used to establish the grading system, it is not unexpected that there was an increase in the number of patients with metastatic evolution according to grade. Therefore, it is necessary to validate this proposed grading system in an independent database and with a larger number of patients. We strongly believe that the application of an objective grading system such as the one described herein for the first time is the first step in providing useful information for the management of atypical Spitz tumors.

Accepted for publication August 21, 1998.

Corresponding author: Raymond L. Barnhill, MD, Division of Dermatopathology and Oral Pathology, the Johns Hopkins Medical Institutions, 600 N Wolfe St, Baltimore, MD 21287-4915.