Dr. Van Voorhees: Do you feel that the diagnosis of melanoma in children is often delayed as a consequence?
Dr. Cordoro: Yes! The atypical clinical features and low index of suspicion contribute to diagnostic delays. Melanomas in children are often treated as warts or pyogenic granulomas, or observed for long periods of time, before the persistence of the lesion or evolution prompts a biopsy. Even then, microscopically ambiguous features, features that do not conform to traditional adult melanoma histopathologic subtypes, or confusion with benign lesions such as Spitz nevi may further delay the diagnosis.
Dr. Van Voorhees: Before your study what did we know about childhood melanomas? Did we have inkling that their clinical features were somewhat different than adult melanomas?Dr. Cordoro: The clinical and histopathologic features of childhood melanoma, especially in pre-pubertal patients, are understudied, and as a result, are poorly characterized. Large cancer databases like SEER and NCDB lack adequate clinical details to determine and describe differences in the clinical presentations of melanoma in pre-pubertal and post-pubertal patients. Several well-done but smaller clinical series have reported the frequent observation that melanomas in young children are more often amelanotic, nodular, and histologically thicker at diagnosis than melanomas in adults. We knew from these previous publications and our own clinical experience that melanomas arising in pre-pubertal patients in particular often fail to conform to traditional ABCDs and are misdiagnosed or diagnosed after a long delay. In general, melanomas arising in teenagers and young adults more often present more typically and in keeping with adult characteristics. In general, though, unless one is very familiar with the small amount of literature on this, I do not think that it is common knowledge among general dermatologists just how common atypical presentations in younger children might be. As a pediatric dermatologist who formerly practiced adult dermatology, I began seriously thinking about this issue during my pediatric derm fellowship in 2007 , when I realized that the literature had significant gaps in this area — thus the study.
Dr. Van Voorhees: Tell us about your study and its findings. What have we now learned?
Dr. Cordoro: The goals of this study were to describe the clinical and histopathological features and outcomes of all cutaneous melanomas diagnosed in children at the University of California, San Francisco, and analyze whether conventional ABCDE criteria (Asymmetry, Border irregularity, Color variegation, Diameter > 6 mm, and Evolution) were adequate to clinically detect them. We also sought to determine whether younger pre-pubertal children present differently from older children and which clinical and histologic variables are associated with metastasis.
Our study highlights key differences in clinical presentation, microscopic features, stage, and outcomes between children aged 0 to 10 years (pre-pubertal) and aged 11 to 19 years. Previous series have identified atypical presentations of pediatric melanoma; however, most lack adequate clinical details and statistical power to draw meaningful conclusions. Our cohort includes 19 pre-pubertal children, representing one of the largest series reporting detailed clinical and histopathological features of melanoma in this age group, and this is clinically very important.
Our study taught us that childhood melanoma often presents unconventionally and current clinical detection parameters in widespread use (ABCD) fail to identify a significant subset of childhood melanomas. Thus, conventional ABCD criteria are inadequate in children. The majority of younger children in our study had melanomas that, in contrast to the traditional ABCD features, presented with amelanosis, symmetry, regular borders, uniform color, and diameters less than 6 mm. The melanomas in older children were evenly split between conventional ABCD (more adult features) and unconventional features. The most sensitive indicator of melanoma in children was a lesion that evolved, thus, the E for Evolution should be universally applicable in all populations, independent of age.
Dr. Van Voorhees: Are the risk factors for melanoma the same as in adults or are they skewed to congenital nevi tumors?
Dr. Cordoro: Some risks are the same, and some are unique to children. The presence of a large congenital melanocytic nevus does present a heightened individual risk to that particular patient, but does not contribute very significantly to overall melanoma risk in children in general. Other risks in children include the presence of numerous (50 or more) melanocytic nevi, dysplastic nevus syndrome, sun-sensitive phenotype, family history of melanoma, and DNA repair defects such as xeroderma pigmentosum. Children who are iatrogenically or genetically immunosuppressed, have received excessive ultraviolet radiation, or have a history of malignancy resulting in radiation and/or immunosuppressive therapies are also at greater risk.
Interestingly, in younger children, darker Fitzpatrick skin types are over-represented; the presence of darker-skinned, non-Caucasian subjects in younger pediatric cohorts suggests that childhood melanoma differs in unique ways from that occurring in adults. Our study confirmed the importance of family history of melanoma in children of all ages and the contribution of numerous acquired melanocytic nevi and personal history of more than three sunburns to risk of melanoma in older children, suggesting the impact of genetic predisposition in younger children and the larger influence of environmental exposures for melanoma development in older children. Only a minority (11 percent) of our cohort developed melanoma in the context of large congenital melanocytic nevi (LCMN), immunosuppression, dysplastic nevus syndrome, or history of cancer.
Importantly, of 322 pediatric patients with melanoma comprising eight single-institution studies, only 22 percent had a predisposing risk factor, underscoring the need for awareness even in the absence of known risk factors.
Dr. Van Voorhees: Does the histology seem the same in children and adults?
Dr. Cordoro: No, especially in the younger children. Pathologically, pediatric melanomas show distinct differences from conventional adult melanoma. Our group includes few cases of melanoma in situ or superficial spreading melanoma. Rather, most presented with nodular, Spitzoid, or unclassifiable melanomas.
These pathologic differences seem to parallel the non-(conventional) ABCD morphology observed clinically.
Interestingly, the majority of melanomas eventuating in death and nearly half (44 percent) overall were not classifiable by experienced dermatopathologists using conventional adult subtypes (e.g., superficial spreading, nodular, acral lentiginous). These categories provide an inadequate structure for the classification of pediatric melanomas. Additional studies are warranted to determine if such melanomas in children can be better stratified by molecular analysis. Sadly, the inapplicability of adult melanoma subtypes to most childhood melanomas also contributes to delayed diagnosis.
Dr. Van Voorhees: Is there a difference in the histology of children depending on their age of onset?
Dr. Cordoro: Yes. Histopathological subtypes in the study population differed significantly between the younger and older groups. Unclassified/ambiguous and Spitzoid melanomas comprised 47 percent of lesions in pre-pubertal children while unclassified melanomas represented the majority in older children. Conventional adult melanoma subtypes such as acral, superficial spreading, and nodular were more prevalent in older children whereas only the nodular subtype was represented in pre-pubertal children.
Dr. Van Voorhees: Is there a difference in risk of metastasis or in survivability?
Dr. Cordoro: Prognostic factors in adult melanoma are well established but are less clear in children. The effect of some factors, such as tumor thickness, may depend on age or pubertal status. In our study, Breslow thickness predicted metastasis but not death, likely as a result of inadequate power to detect an association because of the small number of deaths. Survival estimates for children of different ages vary depending on the source and it remains unclear if outcomes differ significantly from adults. Similar to previous studies, we observed thicker primary lesions and more frequent sentinel lymph node (SLN) metastases in young children compared with adults. To date, the prognostic impact of SLN positivity in patients with pediatric melanoma or ambiguous lesions is unclear, and additional studies from larger cohorts of patients will be required to definitively address this question.
Our cohort revealed more histologically aggressive features (ulceration, mitosis) in younger compared with older children and 92 percent of younger kids presented with stage IIA or higher versus less than half in stage IIA or higher in the older group. Paradoxically, only one death occurred in the younger group, and it was in the context of a LCMN, supporting a less aggressive biologic behavior of Spitzoid and other melanomas occurring in young children while older, presumably post-pubertal, children’s tumors seem to behave more similarly to adult melanoma.
In scrutinizing the features of the 10 patients who died, several features stand out. Diagnosis was delayed for more than one year in 60 percent of patients and 60 percent of patients had at least one risk factor for melanoma. Six of 10 who died had SLN biopsy at diagnosis, and only two specimens revealed positive findings. All three patients with melanoma arising in LCMN died — one infant and two teenagers — with widespread metastases and death within eight months from diagnosis. Interestingly, one of the melanomas arising from a LCMN was amelanotic. In contrast to an emphasis on cutaneous melanoma risk in LCMN in the first decade, these data underscore that melanoma can arise at any age in this setting.
Other risk factors such as prior malignancy, immunocompromise, and dysplastic nevus syndrome may carry increased mortality risk compared with melanomas in children without such risk factors. In our study, seven of 10 melanomas leading to death were clinically amelanotic, raising the unanswered question of whether this subtype carries a more dismal prognosis.
Dr. Van Voorhees: What would you recommend to the practicing dermatologist who is seeing children in his or her office for evaluation of moles — is there any guidance that you’d like to provide?
Dr. Cordoro: It is clear that conventional ABCD criteria for clinical detection of melanoma are not fully adequate for children. Overall, the conventional ABCD criteria captured less than half of the clinical characteristics of melanoma in the younger children, and less than 60 percent of those observed in the older children. The criterion of evolution was universally valuable, capturing nearly 100 percent of the entire cohort. Based on these findings, we propose the following criteria for children be considered when doing skin checks in children:
A = Amelanotic
B = Bleeding, Bump
C = Color uniformity
D = De novo, any Diameter
The goal of the modified ABCD criteria is to raise awareness of the alternate presentations of melanoma in children, not to provide absolute diagnostic criteria or to reject the original ABCDEs the features of which may also be found in childhood melanomas of all ages. Considering all of these characteristics may facilitate earlier detection by modifying the overall concept of how melanoma may present in children. When performing “mole checks” in children, maintain a keen suspicion of any new, persistent (beyond several weeks), or recently changed lump or bump, pigmented or not, with or without asymmetry, irregular border, color variegation, or size greater than 6mm. Children with suspicious lesions even without well-recognized risk factors should be treated with an elevated baseline level of awareness.
Because benign and transient inflammatory lesions such as insect bite reactions and folliculitis are common in children, observing suspicious amelanotic or papulonodular lesions for several weeks before biopsy is appropriate.