|Home | About | Journals | Submit | Contact Us | Français|
Congenital melanocytic nevi (CMN) are pigmented birthmarks that affect up to 80% of the skin surface area. The increased frequency of CMN in families of severely affected individuals is suggestive of a predisposing germline genotype. We noted a high prevalence of red hair in affected families, and considered a role for MC1R in this condition. A cohort of 166 CMN subjects underwent pigmentary phenotyping, with MC1R genotyping in 113. Results were compared with a local control group of 60 unrelated children and with 300 UK children without CMN. CMN subjects had higher prevalences of red hair and a red-haired parent than local controls and had a higher rate of compound heterozygosity and homozygosity for MC1R variants. The presence of a V92M or R allele (D84E, R151C, R160W, D294H) was associated with increasing size of the CMN, implying a growth-promoting effect of these alleles. Unexpectedly, the V92M and R151C alleles were also strongly associated with birth weight in the CMN cohort, a finding confirmed in the control group. The effect of germline MC1R genotype on development and severity of CMN led us to investigate potential broader effects on growth, revealing a role for MC1R in normal fetal development.
Congenital melanocytic nevi (CMN) are congenital moles that cover up to 80% of the body surface area and can number hundreds in a single individual. They are permanent and grow in proportion to the child, although they often lighten in color (Strauss and Newton Bishop, 2008; Kinsler and Bulstrode, 2009; Kinsler et al., 2009b). They are currently categorized by projected adult size (PAS) of the largest nevus and by total number of nevi (Ruiz-Maldonado, 2004). Neurological abnormalities such as neuromelanosis, congenital malformations, hydrocephalus, and benign tumors are the commonest complication of CMN, with the prevalence increasing with the severity of cutaneous lesions (Kadonaga and Frieden, 1991; DeDavid et al., 1996; Kinsler et al., 2008). Primary malignant melanoma of the skin or the CNS is a much rarer complication in childhood (Hale et al., 2005; Krengel et al., 2006; Kinsler et al., 2009a), although almost universally fatal, and prophylactic removal of skin lesions has not been shown to decrease the risk of malignancy (Kinsler and Bulstrode, 2009). The recent finding of characteristic facies in children with CMN has broadened the description of this condition, and the term CMN syndrome has been proposed with preliminary diagnostic criteria (Kinsler et al., Am J Med Genet, 2012 (in press)). The genetic cause of these syndromic features is likely to be a somatic change affecting the ectoderm or neural crest, but is still to be elucidated.
There is, however, evidence of a germline predisposition for the development of CMN syndrome; this includes a higher prevalence of smaller CMN in family members of affected individuals (Kinsler et al., 2009a), together with a handful of reports of large CMN in first-degree family members (Danarti et al., 2003; de Wijn et al., 2010). We hypothesized that the melanocortin-1-receptor (MC1R) gene might have a role in this predisposition, on the basis of a clinical observation of a high prevalence of red hair in families of children with CMN (for examples see Figure 1). The red hair/fair skin/freckling phenotype is strongly associated with variants of MC1R on chromosome 16q24.3 (Valverde et al., 1995; Box et al., 1997; Schioth et al., 1999; Healy et al., 2000; Bastiaens et al., 2001). MC1R is a G-protein-coupled receptor expressed on melanocytes, melanoma cells, and CMN cells (Ghanem et al., 1988; Donatien et al., 1992; Mountjoy et al., 1992; Loir et al., 1999). The gene is highly polymorphic, with over 100 variants reported to date. Variants with a stronger association with the red-hair phenotype in humans have been designated R alleles (D84E, R151C, R160W, and D294H), whereas those with a weaker association are designated r alleles (Sturm et al., 2003). This study was undertaken to examine whether the MC1R genotype influenced the development of CMN.
A total of 166 children with CMN were included in the study. The cutaneous phenotype was classified as in Table 1.
Comparisons between affected and unaffected subjects were restricted to Caucasian subjects of Northern European ancestry (i.e., white-skinned individuals with four Northern European grandparents) who constituted the majority of the study group. Two control groups were ascertained; these were designated as control groups A (60 unrelated children from the same school, street, or locality as the affected individuals) and B (300 UK children without CMN from the Avon Longitudinal Study of Parents and Children (ALSPAC)) (Jones et al., 2000; Golding et al., 2001). Comparison of the pigmentary phenotype of the Northern European white children showed no statistical differences in phenotype between control groups A and B, although as the numbers in group A are relatively small we cannot exclude the possibility that these groups were different in some respect, and as a result we have used control group B for genotype data. There were significantly fewer blonde/light brown–haired individuals in the CMN cohort than in control group A and separately in control group B (P=0.0215 and P=0.0275 respectively; Table 2). In addition, fewer dark brown/black-haired individuals were present in the CMN cohort than in control group A (P=0.0293), whereas this was similar in the CMN cohort and control group B. Conversely, CMN children were more likely to have red hair than the larger control group B (P=0.0085), whereas this was not significantly different from the number of red-haired children in control group A. However, CMN children were more likely to have at least one red-haired parent (32 of 107 (30%)) than control group A (1 of 47 (2%)), respectively (P<0.0001), evenly divided between mothers and fathers. Parental pigmentary phenotype was not available for control group B. The pigmentary phenotype of the Northern European white children is shown in Table 2.
MC1R variants, particularly the R alleles, are known to be causally associated with red hair (Valverde et al., 1995; Box et al., 1997; Smith et al., 1998; Sulem et al., 2007), and there is some evidence that the MC1R genotype may be associated with blonde/light brown hair (Box et al., 1997; Smith et al., 1998). On the basis of these phenotypic observations between affected and unaffected groups, we hypothesized that the MC1R genotype may influence the development of CMN.
According to the phenotyping described above, comparison of the MC1R gene between cohorts was restricted to white-skinned children of Northern European ancestry. To ensure sufficient numbers for statistical analysis of prevalence of MC1R polymorphisms, members of the affected group and control group B were sequenced for MC1R. Individuals with CMN were more likely to have two MC1R variant alleles than controls (P=0.0052, Table 3). More detailed analysis indicated that this was not due to a difference in the frequency of a particular MC1R variant between these two groups (see Supplementary table S1 online). The MC1R genotype in three pairs of affected and unaffected skin samples was identical with the germline genotypes, suggesting that somatic MC1R mosaicism was not the causative feature of affected versus unaffected skin.
In addition to coding region variants, a known SNP in the 3′ UTR (rs3212369) was found at a high frequency in both populations, and found to cosegregate strongly but not exclusively with the V92M and T314T alleles (P<0.001 for Pearson's correlation coefficients on all combinations of two variables). However, as an extension to previous reports (Harding et al., 2000), the pattern of segregation suggested that this SNP arose independently of the T314T variant. This implies that either the V92M variant arose twice independently, on the 3′ UTR (rs3212369) and T314T alleles, or that there was intragenic recombination within MC1R.
In the CMN cohort, the presence of a V92M or an R allele (D84E, R151C, R160W, D294H) was found to be significantly associated with increasing projected adult size (PAS) of the largest CMN (P=0.01) (Figure 2). We examined this effect in more detail using >60cm PAS as a cutoff, as this has previously been noted to be a significant predictor for increased incidence of malignant melanoma (Demenais et al., 2010). The odds ratio for having a CMN of PAS >60cm compared with those of <60cm if carrying V92M or an R allele was 2.867 (95% CI 1.1257–6.537, P=0.012).
Thirty-four percent of the children with CMN had a first- or second-degree family history of CMN. If the MC1R genotype were one of the predisposing factors for development of CMN, a difference in MC1R genotype between those with and without a positive family history of CMN would be expected. This was indeed the case, with a significantly higher prevalence of at least one MC1R variant in those with a family history of CMN versus those in whom no first- or second-degree relative had a CMN (91 vs. 59% respectively, P=0.0062). When confined to the V92M and/or R alleles, a similar preponderance of variants was noted in CMN children with a family history versus those without a family history of CMN (65 vs. 30%, P=0.018).
The above results, in which MC1R variants were associated with larger CMN and more nevi, indicate that the MC1R genotype influences the development of CMN. However, an additional possibility is the existence of a broader role for MC1R genotype in fetal growth and development, especially that MC1R variants function in promoting growth of the fetus. Therefore, we examined for associations between V92M and R alleles and birth weight in the CMN cohort and control group B. The presence of a V92M or R151C was significantly associated with increased birth weight in the CMN cohort (unstandardized coefficient +697g, SE 216g, P=0.002), independent of gender and maternal smoking (Figure 3a). A similar significant effect of the same MC1R alleles was observed on birth weight (+187g, SE 59g, P=0.002) in the ALSPAC cohort, control group B (Figure 3b). This effect was independent of sex, gestation, maternal smoking, maternal weight pre-pregnancy, and maternal hypertension. Thus, the magnitude of this genotype effect is comparable to the effect of gender in this study. There was no significant association between R alleles as a group and birth weight in either cohort. No association between postnatal weight at the age of 49 months and MC1R genotype was found in control group B.
Previous studies of the effects of MC1R genotype in postnatal life have documented fewer acquired melanocytic nevi in red-haired children and adults (Duffy et al., 2004; Dellavalle et al., 2005), and in individuals homozygous or compound heterozygous for R alleles at the MC1R locus (Duffy et al., 2004). In addition, genome-wide association studies have not found any association between MC1R genotype and numbers of acquired melanocytic nevi (Falchi et al., 2006; Yeh and Bastian, 2009), and no association between MC1R variants and higher numbers of acquired nevi in families with inherited melanoma (Demenais et al., 2010). However, in the current study, we have identified associations between families with CMN-affected children and the red hair pigmentary phenotype, and associations between MC1R variants and the presence of CMN, as well with more extensive CMN. Taken in the context of the effect on birth weight, this difference between congenital and acquired nevi may be due to the differential effects of MC1R during pre- and postnatal life, although other inherent genetic differences between congenital and acquired nevi could also be significant factors.
Our findings indicate that certain MC1R variants function to promote the growth of CMN during embryogenesis, and possibly the growth of the fetus as a whole. The exact mechanism whereby MC1R variants alter fetal growth and development are unknown, but might involve direct effects on the fetus and/or indirect effects through the presence of MC1R on trophoblast cells of the placenta (Thornwall et al., 1997). MC1R is expressed on a wide range of adult human cell types including adipocytes (Hoch et al., 2007; www.genecards.com; Rebhan et al., 1997), supporting the increasing body of evidence that it has significant non-pigmentary roles (Robinson and Healy, 2002; Mogil et al., 2005; Robinson et al., 2010); alpha-MSH is involved in metabolic pathways, but this is currently understood to be via the other melanocortin receptors. Melanocortin-1-receptor (MC1R) genotype is known to interact with pigmentary and non-pigmentary pathways (Robinson and Healy, 2002; Cooper et al., 2005; Robinson et al., 2010), for example, altering the penetrance of other genes in melanoma such as p53 (Patton et al., 2005; Stefanaki et al., 2007; Nan et al., 2008) and CDKN2A (Box et al., 2001; Debniak et al., 2006). The latter effects are dependent not only on particular alleles but also on the number of variant alleles present (Demenais et al., 2010). A similar interactive effect may be operating in individuals with CMN, with MC1R variants exerting a phenotype-exacerbating effect on a different somatic, or possibly germline, genetic alteration that affects CMN development. The effect of MC1R genotype on birth weight in our study was a generalized phenomenon that was found in two separate cohorts (CMN and ALSPAC), and in the MC1R-pigmentary context the previously reported interactive effects of MC1R genotype on oculocutaneous albinism (King et al., 2003) was a generalized, rather than localized, effect on pigmentation. The fact that in the ALSPAC cohort of healthy children the larger birth weight in subjects with MC1R variants was not associated with any other dysmorphic features suggests that in the normal situation MC1R genotype may influence fetal development to affect body size in a proportionate manner. Interactive effects of MC1R with other genes affecting birth weight have not yet been investigated, and further work on this topic will be required in the future.
MC1R variants are a risk factor for melanoma development in the general population and are associated with increased risk of melanoma in CDKN2A mutation carriers (Box et al., 2001; Demenais et al., 2010). One important clinical consideration in our patient cohort is whether the pattern of germline MC1R genotype is responsible for the excess risk of malignant melanoma in individuals with CMN. The lifetime risk for melanoma is around 1–2% when all CMN are considered together as a group (Krengel et al., 2006; Kinsler et al., 2009a), which is similar in magnitude to the excess risk associated with MC1R variants in the normal population. The numbers of melanomas in our CMN cohort are too few to allow detailed investigations on the risk of melanoma attributable to MC1R variants, but it is likely that this genotype is at least contributing to melanoma risk. In a recent family-based melanoma study, the odds ratio for melanoma in CDKN2A mutation carriers was 5.67 (95% CI 2.1–15.29) for subjects with MC1R variants (Demenais et al., 2010); thus, it is possible that the MC1R genotype may similarly alter the risk of melanoma in people with larger CMNs.
In conclusion, we have identified associations between CMN and red-hair/MC1R variants. Furthermore, the presence of a V92M or R allele was associated with larger CMN size and with larger birth weight in the CMN, and separately in the ALSPAC, groups consistent with a growth-promoting effect of these alleles during embryogenesis.
This study was approved by the research ethics committee of Great Ormond Street Hospital for Children/Institute of Child Health, London. All aspects of this study comply with the Declaration of Helsinki Principles. Written, informed consent was given by parents in all cases, and by subjects, where appropriate.
In all, 166 families with children with CMN were recruited for pigmentary phenotyping during clinical consultations, and offered MC1R genotyping. CMN phenotype was classified as previously published (Ruiz-Maldonado, 2004; Kinsler et al., 2008, 2009a, 2009b) (Table 1; for examples see Figure 1). The total number of nevi at birth was ascertained by history, direct examination, or photographs. Mean and median ages at enrollment to the study were 5.02 years (SEM 0.53) and 2.54 years (range 0.03–43.12), respectively. Follow-up during the study period was for a mean of 3.09 years and median of 3.12 years. Pigmentary phenotyping was performed by a single observer (VK) and was assessed after 1 year of age because of the changes frequently seen in hair and eye color in the first year. Hair color was classified using a chart of 12 colors; eye color was grouped into six standard colors, with peri-pupillary and predominant background iris color considered separately; freckling was recorded by grouped number and distribution; and skin type was assessed using the Fitzpatrick score. Pigmentary phenotype was obtained for 148 children, 143 mothers, and 130 fathers. Ancestry, up to the four grandparents, was also recorded; patients were classified as white Northern European Caucasians if their skin color was white and if all four of their grandparents were born in Northern Europe. Consent was obtained for MC1R genotyping in 113 families. Adequate blood samples for DNA extraction were obtained from 104 children, 78 mothers, and 58 fathers. With further written consent, three children had skin biopsies taken from affected and unaffected skin, whereas they were undergoing general anesthesia for other procedures.
With the assistance of the CMN families, we recruited a geographically matched and sex-matched control population of individuals without CMN. Sixty families completed the same pigmentary phenotyping questionnaires as the CMN group, with telephone guidance from the same researcher. Fifty-five were white Northern European Caucasians, i.e., four grandparents born in Northern Europe; the mean and median age of these 55 children was 7.2 years (SEM 1.1) and 5.6 years (range 0.7–37.6), respectively. To check whether self-reported phenotypes were accurate, 25 CMN families (66 individuals) completed the pigmentary phenotype questionnaire before undergoing face-to-face phenotyping. Pearson's correlations for hair color and eye color were 0.509 and 0.855, respectively, significant at <0.05 level, supporting the use of self-reported questionnaires in this context.
Ethical approval for this part of the study was obtained from the ALSPAC Law and Ethics Committee and the Local Research Ethics Committees. The children in this group born in 1992 had face-to-face pigmentary phenotyping by trained observers at the age of 49 months, and the presence of CMN was noted. The 300 samples were chosen randomly from those without CMN. Pigmentary phenotyping in the ALSPAC study was performed by trained research nurses as follows: hair color was compared with 11 real-hair swatches of measured luminosities, eye color was compared with six standard iris photographs, and skin reflectance was measured by spectrophotometer. MC1R genotyping was performed for all 300 samples, with 270 reliable results for Northern European Caucasian white children.
DNA was extracted directly from all samples using standard extraction procedures. The 951 coding base pairs of the single exon of MC1R, as well as a further 416 base pairs of the 3′ UTR, were amplified by PCR as a single product and sequenced using four overlapping primer sets (5′-TCCAGCCAGGGAGAGGGTGTGA-3′ and 5′-AGCTGCAGGTGATCACGTCAAT-3′ 5′-CTCACCCATGTACTGCTTCA-3′ and 5′-GAGTGTGAGATGCAGGAAGAAG-3′ 5′-CTCATGGCCGTGCTGTACGT-3′ and 5′-CTTTGACAAACGGGGACCAG-3′ 5′-ACGCTCAAGGAGGTGCTGACAT-3′ and 5′-CCACCCAGAGACTTCACAT-3′).
In the CMN cohort, data on birth weight and maternal smoking were available for 128 children. Pre-term children were excluded on the basis that there might have been other abnormalities during the pregnancy that could confound any genotype effect. In control population B, data were available for birth weight, gestation, maternal smoking, maternal weight before pregnancy, and maternal hypertension for all children.
Comparisons of prevalence between groups (pigmentary characteristics and MC1R genotype) were made by Fisher's exact test. Analysis of the statistical effect of genotype on CMN phenotype and on birth weight was by ordinal and linear logistic regression analysis, respectively. For the MC1R genotype associations, we tested the following four variables initially against projected adult size of the main nevus: V92M, V60L, R163Q, and R alleles (grouped together, as they are known to be a clinically and physiologically cohesive group, but some are rare alleles); other alleles were excluded from this analysis as they were found in <5% of our population. To improve our model fit, we then combined the V92M and R allele groups to make the variable V92M or R allele. A Bonferroni correction was applied, reducing the level of significance from 0.05 to less than 0.0102062 (i.e., for a total of five statistical tests). We then used this combined variable, V92M or R allele, to test against birth weight in the CMN cohort, with a further refinement made to the model by using V92M combined with R151C (the commonest of the R alleles), to make the variable V92M or R151C. Again, with a Bonferroni correction (for two tests), this reduced the level of significance (P-value) from 0.05 to 0.0253206. For genotype testing against birth weight in the control group, we tested the single variable V92M or R151C.
This research was funded by the Wellcome Trust, and the Caring Matters Now charity. For the CMN cohort, we are extremely grateful to all the families who took part in the study. For the ALSPAC cohort, we are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses. The UK Medical Research Council (grant ref: 74882), the Wellcome Trust (grant ref: 076467), and the University of Bristol provide core support for ALSPAC.
The authors state no conflict of interest.
Supplementary material is linked to the online version of the paper at http://www.nature.com/jid