Classification of Neurological Abnormalities in Children With Congenital Melanocytic Naevus Syndrome Identifies Magnetic Resonance Imaging as the Best Predictor of Clinical Outcome

R. Waelchli; S.E. Aylett; D. Atherton; D.J. Thompson; W.K. Chong; V.A. Kinsler


The British Journal of Dermatology. 2015;173(3):739-750. 

In This Article


In total, 289 children fulfilled the criteria for MRI of the whole CNS. Two hundred and seventy-one of these scans were performed successfully. The mean and median ages at time of first MRI were 1·5 and 0·6 years, respectively (SEM 0·2), and the mean and median length of follow-up was 11·0 and 8·5 years, respectively (SEM 0·4). The cutaneous phenotype spectrum of this tertiary referral centre cohort is skewed towards the severe end, with 62% of the patients having a CMN of > 20 cm projected adult size, and 41% of > 40 cm (Table 1).

Overall, 46 of 271 (17%) MRIs from the whole cohort were abnormal, rising to 21% in the post-2008 criteria cohort (nonsignificant difference; see below for comparison of pre- and post-2008 data). Abnormalities were subclassified into group 1, 'intraparenchymal melanosis alone' (n = 28), as this is the most common single finding in patients with CMN, and group 2, 'other CNS pathology' (n = 18). Group 2 was too small to subdivide for analysis, as it included a wide variety of different CNS pathologies (Table 2).

Any radiological abnormality (as a binary variable, normal vs. abnormal MRI) was a stronger predictor than projected adult size for all outcome measures: seizures [odds ratio (OR) 13·4, 95% confidence interval (CI) 4·7–38·2], neurodevelopmental problems (OR 3·0, 95% CI 1·3–7·0) and requirement for neurosurgery (OR 71·0, 95% CI 8·9–567·3) (Table 3). CNS melanoma and death were not modelled as the numbers are low. When projected adult size of the largest CMN was included in these models as a categorical variable (< 10, 10–20, 20–40, 40–60 and > 60 cm), the variable as a whole was only significant for neurodevelopmental abnormalities (P = 0·03). Sex (as a binary variable, male = 1) was not found to be significant in these models. For the full regression analysis, please see Appendix S1 (Supporting Information).

Seizures and abnormal neurodevelopment were seen in a minority of the normal MRI group; however, not only were these numbers small, but the seizures were also a temporary problem and/or easy to control with a single medication, and the neurodevelopmental abnormalities were mild, compatible with normal schooling. This is a notably milder clinical phenotype than that seen in those with seizures or neurodevelopmental abnormalities in the other radiological groups (Table 3). Of note, one patient with a normal initial MRI scan developed primary CNS melanoma requiring neurosurgery and later died from primary CNS disease (Fig. 2). This patient was first scanned in the late 1990s and the quality of MRI has improved since then. Therefore, it is possible that there was a neurological disease present that was below the resolution of scanning, as has been shown histologically.[13]

Figure 2.

(a) Comparison of the percentage of patients in whom a magnetic resonance imaging (MRI) scan was performed (white column) and not performed (black column) before and after 2008. By excluding those with only a single congenital melanocytic naevus (CMN), independent of size or site, the introduction of guidelines in 2008 has significantly reduced the percentage of patients scanned routinely. However, the percentage of abnormal scans is not significantly altered, suggesting that we have become more efficient at detecting the same rate of abnormalities. (b) Comparison of the percentage of patients with a normal MRI result (white column) with those with an abnormal result (black column) before and after 2008. The introduction of guidelines in 2008 has not significantly altered the percentage of abnormal scans detected, which implies that we are not failing to detect significant numbers of abnormalities. (c) Subclassification of the radiological abnormalities in this cohort of children with CMN and correlation with the incidence of the different clinical outcome measures in each group. CNS, central nervous system.

Of those with abnormal scans, there was a striking difference between the two subgroups in clinical outcomes. While a substantial proportion of patients in group 1 had neurological symptoms, namely 25·9% with seizures, and 28·6% with neurodevelopmental abnormalities (not significantly different from group 2), there were no patients with melanoma in this subgroup, and no deaths. Although CNS melanoma must be possible in this group the risk appears to be low in childhood. Furthermore, we can conclude that in patients with intraparenchymal melanosis alone, symptoms can be related to that congenital disease and do not necessarily equate with malignancy or death. Therefore, with classical radiological features intraparenchymal melanosis alone does not require surgical intervention. Because of this subtlety – the differentiation between symptomatic congenital disease and new onset of symptoms from CNS melanoma – we would continue to recommend a repeat MRI of the CNS in any child who presents with new neurological symptoms. This can then be compared with baseline scans to look for new lesions or progression.

The clinical outcome pattern in group 1 was different to that of group 2, where both the requirement for neurosurgery and the mortality rate from CNS melanoma was substantial (Table 3). The requirement for neurosurgery in group 2 was significantly higher compared with group 1 (Fisher's exact test P < 0·01), but the mortality rate from CNS melanoma did not differ, perhaps owing to low numbers (Fisher's exact P = 0·06). However, group 2 was very heterogeneous. The clinical symptoms in this subgroup ranged from none to mild speech delay to severe global delay (Table 2). Three of the 18 patients presented with an isolated benign nonmelanotic CNS tumour (patients 6, 9 and 13; see Table 2), all of which were removed by neurosurgery with no further sequelae. Others in this group had leptomeningeal disease, and while in four cases this was a progressive malignant process resulting in death, it is important to note that five patients have stable nonprogressive leptomeningeal disease (patients 3, 7, 8, 11 and 14; see Table 2), all involving focal lesions of unknown histology, which are, in some cases, quite extensive (Fig. 3). In all of these cases the lesions are currently asymptomatic. Furthermore, in patient 5 (aged 10 years at the time of writing) there is diffuse leptomeningeal disease indistinguishable radiologically from the four patients with progressive malignant process, which developed over the first year of life but which has not changed since then. Therefore, patients in group 2 have to be assessed on an individual basis, and our experience suggests that radiological progression or stability are useful guides to management. This is particularly useful in a condition where histology is often not reliable. New genetic tests on leptomeningeal lesions are likely to be helpful in the future to distinguish between benign and malignant lesions.

Figure 3.

(a) Intradural extramedullary disease, presumed leptomeningeal, dorsal to the spinal cord. (b) Enhancing intradural extramedullary disease, presumed leptomeningeal, dorsal to the spinal cord. (c) Intradural extramedullary disease, presumed leptomeningeal, dorsal to the spinal cord. (d) Prominent central canal of the lower thoracic cord. These lesions have not been biopsied and therefore no exact diagnosis is available. However, in all cases shown the patients do not exhibit spinal symptoms or signs, and magnetic resonance imaging (MRI) appearances have been stable since birth, with follow-up now at ages 5, 6, 10 and 16 years. (e) Extensive intraparenchymal melanosis of the cerebellum and focal melanosis of left thalamus. (f) Pre- and postgadolinium-enhanced MRI showing meningioma in left sylvian fissure. (g) Pre- and postgadolinium-enhanced MRI showing Dandy–Walker malformation with congenital leptomeningeal disease at 13 days. Further progression of leptomeningeal disease at 22 months. Intraparenchymal melanoma of left sylvian fissure at the age of 24 months, which appears to have developed from leptomeningeal infiltration, although this is not always the case.

Comparison of our practice and results before and after the publication of guidelines on MRI in 2008 identified that we are doing significantly fewer scans as a proportion of new patient referrals since these guidelines were introduced (Fig. 2), whereas the detection rate of abnormalities in those having scans has not significantly changed (small increase) (Fig. 2). Importantly, the clinical phenotype profile of the cohort has not changed since the guidelines were changed. Therefore, we are confident that on the basis of this large prospective dataset the 2008 guidelines are fit for purpose, allowing detection of neurological abnormalities without performing unnecessary scans in very low-risk individuals. However, as with all guidelines, there must always be room for clinical judgement.