Melanoma Arising in Association With Blue Nevus

A Clinical and Pathologic Study of 24 Cases and Comprehensive Review of the Literature

Sanam Loghavi; Jonathan L Curry; Carlos A Torres-Cabala; Doina Ivan; Keyur P Patel; Meenakshi Mehrotra; Roland Bassett; Victor G Prieto; Michael T Tetzlaff


Mod Pathol. 2014;27(11):1468-1478. 

In This Article


In the current study, we describe the clinical and histopathologic features of 24 melanomas arising in association with common or cellular blue nevi and correlate these features with annotated clinical follow-up information. Ours is the first study to demonstrate a significant association between the variable of tumor thickness (defined as either the standard Breslow thickness or, if not available due to lack of orientation or lack of epidermis, the largest dimension) with both recurrence-free survival and reduced time to distant metastasis; other prognostic indicators predictive of outcome in conventional cutaneous melanomas did not correlate with outcome in these rare and unique melanomas.

Since the first description of melanomas arising in association with or mimicking blue nevi, there has been controversy regarding whether clinical and/or histopathologic prognostic indicators (age, gender, Breslow thickness, mitotic figures, and ulceration) that are predictive of clinical outcome in conventional cutaneous melanomas might also apply to this group of melanomas. For example, because the vast majority of standard cutaneous melanomas arise in association with an intraepidermal component, the Breslow thickness (tumor depth from the anatomic boundary of the epidermis) is accepted (and proven) to accurately reflect a given lesion's intrinsic propensity for clinically aggressive behavior.[57] In contrast, the tendency of melanoma blue nevus type to arise as nodules—either in the dermis or in the subcutaneous tissue without an associated intraepidermal component and thus, without a true epidermal origin—calls into question whether a measurement of Breslow thickness accurately captures such a lesion's true propensity for biologically aggressive/invasive behavior. Further complicating this issue is the relative paucity of cases described in the literature and the lack of consensus regarding the diagnostic criteria.[5] In a study by Martin et al,[7] the authors described the clinical and histopathologic characteristics of 23 'blue nevus-like melanomas' and demonstrated that these lesions do not exhibit differences in survival or an increased risk of lymph node or distant metastasis when compared with conventional melanomas matched for age, sex, Breslow thickness, Clark level, ulceration, and anatomic site. Moreover, they demonstrated a statistically significant difference (P=0.002) between the mean Breslow thickness of lesions in patients who developed locally recurrent disease (mean Breslow thickness=9.4 mm) and those who did not (mean Breslow thickness=4.2 mm), while other histopathologic features were not predictive of outcome in these lesions.[7]

Our findings underscore and expand upon these findings. Namely, we show that increasing tumor thickness (which we defined as either the Breslow thickness, when applicable, or the largest tumor dimension available for lesions growing as more deeply situated nodules) correlates with recurrence-free survival (hazard ratio=1.02 per mm; P=0.04) and reduced time to distant metastasis (hazard ratio=1.03 per mm; P=0.02) with a similar trend toward reduced time to local recurrence (hazard ratio=1.02 per mm; P=0.07). Taken together with the results of Martin et al,[7] our findings argue that the largest tumor dimension is an important variable to capture in the synoptic report—particularly for deeply situated lesions of this sort. We did not observe a statistically significant relationship between other parameters (including age, gender, mitotic figures, tumor necrosis, and lymphovascular invasion) and clinical outcome in our series. An additional interesting observation in our study was the finding that despite the relatively high mitotic figure count (mean 6.5/mm2 for all cases) and the relatively large size of these lesions (mean tumor thickness: 20.9 mm), ulceration was a relatively rare event compared with conventional nodular melanomas with similar size and mitotic figure count.[58–61] This observation may be attributable to the fact that many of our lesions arose in the deep dermis/subcutis, supporting the concept that ulceration is likely related to the relationship/proximity of the lesion to the epidermis. Alternatively, recent gene expression profiling studies have shown that ulcerated primary melanomas have distinct gene expression profiles compared with their non-ulcerated counterparts, suggesting that ulceration may in fact be driven by properties intrinsic to the tumor.[62]

To further our understanding of melanomas arising in association with or mimicking blue nevi, we performed a comprehensive review of the literature to systematically classify the clinical–pathologic characteristics of these lesions and to contextualize the cases we describe. We identified 109 melanoma blue nevus type described between 1953 and 2012 in the English literature either as individual case reports or in case series, further underscoring the relatively rare nature of this entity (Figure 3).[4,5,7,11–54] Overall, patient age ranged from 3 to 77 years[5,48] with a median age of 44.7 years similar to what we report here (range: 20–85 years; median 50 years; Figure 3b). Overall, there is a slight male preponderance described in the literature (56 lesions in males compared with 51 women)—in contrast to our findings (9 men to 15 women). The most common anatomic site for these lesions is the head and neck region in 52 cases,[4,5,7,11,13,15,21,23,24,27,28,31,33,36,39,42,46–48,51,53] followed by the trunk in 28 cases,[4,5,7,11,13,16,18,25,32,34,35,38,43,45,50,52] the lower extremity in 11 cases,[4,7,13,14,26,30,33] the upper extremity in 8 cases,[7,13,18,29,40] the buttock region in 7 cases,[4,5,16,33,37] the perineal region in 2 cases,[22,49] and the uterine cervix in 1 case.[41] We observed a similar anatomic distribution in our cohort of cases (Table 1); however, none of our cases occurred on the extremities—a finding identical to what Granter et al[5] reported in their series. This significant predilection for the head and neck supports our standard recommendation that incompletely excised/sampled melanocytic lesions with any features of blue nevus arising in this anatomic site that extensively involve the tissue edges warrant at least conservative re-excision to ensure complete removal and to completely evaluate these lesions—particularly those lesions with a reported history of changing appearance (Figure 1).

Figure 3.

Associated nevus component (a), age (b), gender (c), and anatomic distribution (d) for all reported cases in the English literature, to date, combined with our data from this study. Not every parameter is provided for each reported case in the literature and therefore, the figures represent cases for which the specific respective parameters were reported. (a) Provides a breakdown of the associated nevus component for our cases as well as those reported in the literature (n=108)4,5,7,11–54 (b) Age (n=111),4,5,7,11–35,37–54 (c) gender (n=131)4,5,7,11–35,37–54 and (d) anatomic distribution (n=135)4,5,7,11–54 for all cases.

In contrast to our findings, in their series of 23 patients Martin et al[7] showed that blue nevus-like melanoma was more common in men and demonstrated an even distribution among body sites (ie no obvious predilection for the head and neck). The median age of patients at diagnosis in their study was similar to what we observed (~5th decade). The reported median tumor thickness in their series (5.5 mm) was less than what we found in our cases (12.0 mm) with the caveat that (1) they used Breslow thickness and we used the greatest tumor dimension when Breslow thickness was not applicable and (2) our cases are derived from the experience at a tertiary cancer referral center that might produce a selection bias for larger lesions that are more difficult to treat and/or exhibit clinically aggressive behavior.

Regarding pathologic findings, the reported cases (in order of frequency) included malignant melanocytic proliferations mimicking a cellular blue nevus but lacking an identifiable benign component (in 38% of cases)[5,11,13–15,18,20,25,28,30,33,36,47,49] or melanomas arising in association with a distinctive blue nevus component, including (a) cellular blue nevus (26%),[4,5,11,16,18,29,40,46,50,51,53] (b) common blue nevus (21%),[5,13,22,24,27,31,32,34,37,39,41,43,52] (c) both cellular and common blue nevus (13%),[12,16,21,23,35,38,44,45,48] (d) neurocristic hamartoma (1%)[42] and (e) nevus of Ota (1%).[17] In contrast, all of the cases in our series consisted of distinctly biphenotypic melanocytic proliferations (melanoma together with a benign blue nevus component). The most common associated blue nevus pattern in our series was both common and cellular blue nevus (nine cases; 38%), followed by cellular blue nevus (seven cases; 29%), and common blue nevus (six cases; 25%). Blue nevus/neurocristic hamartoma (one case; 4%) was less common. In one case in our series, melanoma arose in association with an atypical cellular blue nevus (4%). Identifying a benign nevus component in lesions like these is an important clinical variable (for staging and subsequent management), as an associated benign nevus comprises strong evidence for a primary lesion, which otherwise might be in debate for a predominantly dermal/subcutaneous-based malignant melanocytic proliferation without an obvious intraepidermal component/radial growth phase. In our experience, one helpful clue that is suggestive of an associated blue nevus component is the presence of a 'dumbbell-shaped' architecture to the melanoma.

An important limitation to our study was the fact that we designed our search for cases such that we specifically captured lesions with an unequivocally benign blue nevus and/or cellular blue nevus component. Therefore, the conclusions drawn from this study may only pertain to lesions with a distinctive benign blue nevus component and thus may not be as applicable to melanomas whose architecture resembles cellular blue nevi but without evidence of a true benign component. In contrast, most of the cases included in one of the larger studies available in the literature by Granter et al[5] consisted of 'malignant cellular blue nevi' without a true benign nevus component. This may in part explain some of the differences observed by Granter et al[5] in comparison with our findings, including differences in observed survival. In particular, all of their patients for whom follow-up information was available either developed local recurrence or metastasis or died of disease, while five (21%) of the patients in our series had no documented evidence of regional or distant metastasis and were alive at the time we concluded our study. Similarly, necrosis was a relatively common feature observed in our cases (46%) while only 20% of their cases demonstrated necrosis.

When we grouped the lesions according to the type of blue nevus component associated with the melanoma (common blue nevus, Supplementary Figure 1A; cellular blue nevus, Supplementary Figure 1B and those lesions resembling a cellular blue nevus, Supplementary Figure 1C, several interesting observations emerged. First, melanomas resembling cellular blue nevus but lacking a true benign component exhibit a slightly higher median age (52 years) compared with those arising in association with or common or cellular blue nevi (46 years and 43 years, respectively). Second, these melanomas exhibit an anatomic predilection for the head and neck independently of their associated blue nevus component. Thus, the anatomic site typical of the respective benign blue nevus counterpart does not impact the favored anatomic site of the so called 'malignant blue nevus'.

Here, we summarize the features of 24 melanomas arising in association with common or cellular blue nevi and correlated these clinical and histopathologic features with annotated clinical follow-up information. We demonstrate a significant association between the variable of tumor thickness (defined as the largest recorded tumor dimension or Breslow thickness when applicable) with both recurrence-free survival and reduced time to distant metastasis, while other prognostic indicators predictive of outcome in conventional cutaneous melanomas do not correlate with outcomes in these rare and unique melanomas. We advocate inclusion of this variable in our synoptic reporting scheme. Additional larger studies are warranted to confirm this observation.