Advertisement

Genes Determining Nevus Count and Dermoscopic Appearance in Australian Melanoma Cases and Controls

Open ArchivePublished:August 15, 2019DOI:https://doi.org/10.1016/j.jid.2019.05.032

      Abbreviations:

      BNMS (Brisbane Nevus Morphology Study), CI (confidence interval), CM (cutaneous melanoma), OR (Odds Ratio), SNP (single nucleotide polymorphism), TNC (total nevus count)
      To the Editor
      Total body nevus counts (TNC) is a highly heritable trait, with twin studies estimating that 60% to 70% of its variance is explained by genetic factors (
      • Lee S.
      • Duffy D.L.
      • McClenahan P.
      • Lee K.J.
      • McEniery E.
      • Burke B.
      • et al.
      Heritability of naevus patterns in an adult twin cohort from the Brisbane Twin Registry: a cross-sectional study.
      ). Polymorphisms within IRF4, MTAP, PLA2G6, and MITF have been shown to strongly influence TNC, and many other nevus-associated genes have been recognized (
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      ). Dermoscopy has allowed the recognition of distinct morphological classes of nevi, including reticular, globular, and homogenous and/or complex, that tightly correlate with histopathological subtypes (
      • Tan J.M.
      • Tom L.N.
      • Soyer H.P.
      • Stark M.S.
      Defining the Molecular Genetics of dermoscopic naevus patterns.
      ). There has been great interest in the clinical presentation (
      • Suh K.Y.
      • Bolognia J.L.
      Signature nevi.
      ), body site density (
      • Bataille V.
      • Bishop J.A.
      • Sasieni P.
      • Swerdlow A.J.
      • Pinney E.
      • Griffiths K.
      • et al.
      Risk of cutaneous melanoma in relation to the numbers, types and sites of naevi: a case-control study.
      ,
      • Blake T.
      • McClenahan P.
      • Duffy D.
      • Schaider H.
      • McEniery E.
      • Soyer H.P.
      Distribution analyses of acquired melanocytic naevi on the trunk.
      ), dermoscopic growth pattern, and anatomical distribution of nevi (
      • Bajaj S.
      • Dusza S.W.
      • Marchetti M.A.
      • Wu X.
      • Fonseca M.
      • Kose K.
      • et al.
      Growth-curve modeling of nevi with a peripheral globular pattern.
      ,
      • Fonseca M.
      • Marchetti M.A.
      • Chung E.
      • Dusza S.W.
      • Burnett M.E.
      • Marghoob A.A.
      • et al.
      Cross-sectional analysis of the dermoscopic patterns and structures of melanocytic naevi on the back and legs of adolescents.
      ,
      • Marchetti M.A.
      • Kiuru M.H.
      • Busam K.J.
      • Marghoob A.A.
      • Scope A.
      • Dusza S.W.
      • et al.
      Melanocytic naevi with globular and reticular dermoscopic patterns display distinct BRAF V600E expression profiles and histopathological patterns.
      ) in relation to the cutaneous melanoma (CM) risk (
      • Ribero S.
      • Zugna D.
      • Osella-Abate S.
      • Glass D.
      • Nathan P.
      • Spector T.
      • et al.
      Prediction of high naevus count in a healthy UK population to estimate melanoma risk.
      ). In this study, we test the association between the number of nevi of each morphology and melanoma risk and whether known nevus loci predispose to particular nevus morphologies in a genome-wide association study of melanoma cases and controls within the Brisbane Nevus Morphology Study (BNMS).
      We captured dermoscopic images for 27,221 nevi ≥ 5mm in diameter from 1,266 individuals, classified by subtype using two methods (
      • McWhirter S.R.
      • Duffy D.L.
      • Lee K.J.
      • Wimberley G.
      • McClenahan P.
      • Ling N.
      • et al.
      Classifying dermoscopic patterns of naevi in a case-control study of melanoma.
      ), which were pooled for analysis (see Supplementary Materials). In the combined dataset, the dominant nevus subtype pattern was nonspecific (homogenous 46% and/or complex 37%), followed by reticular (11%), and then globular pattern (5%). On average, the number of nevi in each subtype was higher in the cases than the controls, with nonspecific at 24.2 versus 9.3, reticular at 5.2 versus 2.2, and globular at 2.7 versus 1.8 (Supplementary Table S1).
      The odds ratio (OR) for CM on the quintiles of the nevus subtype counts, compared with the lowest quintile (0% to 20% set to 1) for each subtype are shown in Figure 1 (Supplementary Table S2). There was a significant increase in each of the four quintile comparisons of nonspecific subtype count, the greatest being in the highest quintile of counts for each subtype, with an OR of 7.64 (95% confidence interval [CI] = 4.86-12.19) for nonspecific nevi. In the subsample where complex and homogenous counts were available, the homogenous count was the most strongly associated phenotype for CM with OR of 11.34 (95% CI = 5.63-23.75) for the top quintile of homogenous nevi; for complex nevi, OR of 4.75 (95% CI = 2.40-9.62) (Supplementary Tables S3 and S4). Despite the reduction of globular subtype counts with age, the OR was 2.82 (95% CI = 1.72-4.88), with reticular subtype count OR of 2.94 (95% CI = 1.84-4.77), also a significant predictor of risk (Supplementary Table S2).
      Figure thumbnail gr1
      Figure 1Odds ratio for cutaneous melanoma by nevus subtypes. CM odds ratio and 95% CI (x-axis) is plotted by quintile (y-axis) of the three nevus subtypes: globular, reticular, and nonspecific. The first quintile set to 1 (0-20%) is not shown. CI, confidence interval; CM, cutaneous malignanat melanoma; Glob, globular; NS, nonspecific; Ret, reticular
      The genome-wide association analysis of melanoma, TNC, and nevus subtypes was performed on 1,235 BNMS participants with single nucleotide polymorphism (SNP) genotype and phenotype data available. Figure 2 shows Manhattan plots of the P-values for the SNPs across the genome (gray) for (i) TNC and (ii) globular nevus counts, overlaid with the SNPs of significance in each peak for 32 melanoma or nevus loci (blue) from our recent genome-wide association study meta-analyses of melanoma and TNC (
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      ,
      • Law M.H.
      • Bishop D.T.
      • Lee J.E.
      • Brossard M.
      • Martin N.G.
      • Moses E.K.
      • et al.
      Genome-wide meta-analysis identifies five new susceptibility loci for cutaneous malignant melanoma.
      ). The SNPs of greatest significance for the TNC in these candidates (Supplementary Table S5) are within the MTAP (P = 2x10-6) and PLA2G6 (P = −1.7x10-3) loci, with the SLC45A2 rs250417 SNP (P = 2x10-5) being protective, but do not in themselves reach genome-wide significance, as expected given the sample size. One SNP in the IRF4 gene rs12203592*C/T (P = 7x10-8) reached a level close to genome-wide significance when globular nevi were considered alone but not for the total nevus count (Figure 2b, Supplementary Table S5). Notably, this polymorphism was first reported to influence TNC in an age-dependent manner (
      • Duffy D.L.
      • Iles M.M.
      • Glass D.
      • Zhu G.
      • Barrett J.H.
      • Höiom V.
      • et al.
      IRF4 variants have age-specific effects on nevus count and predispose to melanoma.
      ), with its genotypic-specific effect on globular nevus counts consistent with the loss of globular nevi with age (Supplementary Figure S1). The SNPs within the MTAP locus were also consistently associated with globular nevus count, but none achieved P < 10-4. For reticular and nonspecific nevus counts (Supplementary Figure S2), the SNPs within the MTAP locus are again the highest of the candidate genes, with the most significant SNP rs7036656 (P = 2.5x10-5) located near the flanking CDKN2A gene.
      Figure thumbnail gr2
      Figure 2Genome-wide association study plot for total nevus count and globular nevus count. Manhattan plot of P-values for (a) TNC ≥ 5mm and (b) globular nevus count ≥ 5mm with genome-wide significance indicated by the black dashed line (5x10-8) and suggestive of significance (5x10-7) by the gray dotted line. Blue dots indicate the SNPs within previously associatied peaks, and the gray dots represent all other SNPs on the Illumina HumanCoreExome array. The peak-associated SNP rs7023329 on chromosome 9 is indicated as MTAP, rs132985 on chromosome 22 as PLA2G6, and rs12203592 on chromosome 6 as IRF4. SNP, single nucleotide polymorphism.
      In the case of melanoma, the most strongly associated candidate genes in the BNMS, compared with a larger number of controls drawn from the QIMR Brisbane Longitudinal Twin Study and other studies (
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      ), were MTAP, PLA2G6, IRF4, FMN1, MC1R, ASIP, and SLC45A2. There were 26 BNMS individuals who were carriers of the MITF E319K variant rs149617956*G/A, which as expected was highly associated with both melanoma and TNC (Supplementary Table S5).
      The goals of the BNMS were to assess the natural history of the dermoscopic morphological subtypes of nevi, examine the relationship of subtype counts to CM, and identify genetic risk factors for CM for which these counts would be the intermediate step in pathogenesis. We found that the number of homogenous subtype nevi was the most highly associated with CM risk, followed by the complex pattern, then reticular and globular. This is consistent with the transition of other dermoscopic patterns to a nonspecific morphology with increasing age (
      • Zalaudek I.
      • Schmid K.
      • Marghoob A.A.
      • Scope A.
      • Manzo M.
      • Moscarella E.
      • et al.
      Frequency of dermoscopic nevus subtypes by age and body site: a cross-sectional study.
      ). Dermoscopy is currently used in melanoma surveillance to identify clinically significant nevi. Based on this study, the number of common acquired nevi with particular dermoscopic patterns may allow a slightly better assessment of melanoma risk. For each additional nonspecific nevus, the risk for melanoma increased by approximately 2%. The combined homogenous and/or complex nevus types comprised 83% of our cohort; as such, this is consistent with the increased risk with increasing TNC. Incorporating globular and reticular numbers in the model increased the melanoma predictive accuracy by approximately 1%.
      In this study, the rs12203592*T allele was strongly associated with the globular dermoscopic nevus pattern, with the homozygous *T/T genotype reducing globular nevus counts. This observation is consistent with the report in the SONIC study of children (with a mean age of 10.4 years) (
      • Orlow I.
      • Satagopan J.M.
      • Berwick M.
      • Enriquez H.L.
      • White K.A.
      • Cheung K.
      • et al.
      Genetic factors associated with naevus count and dermoscopic patterns: preliminary results from the Study of Nevi in Children (SONIC).
      ) where the rs12203592*T allele was associated with a lower probability of having a globular morphology (than the homogenous pattern), even though it increased the total nevus count. As reported previously, adult *T/T carriers have reduced TNC, the reverse to that observed in children where *T/T have higher TNC (
      • Duffy D.L.
      • Iles M.M.
      • Glass D.
      • Zhu G.
      • Barrett J.H.
      • Höiom V.
      • et al.
      IRF4 variants have age-specific effects on nevus count and predispose to melanoma.
      ).

      Ethics Statements

      This study was approved by the Metro South Human Research Ethics Committee (approval #HREC/09/QPAH/162, 26 August 2009) and The University of Queensland (approval #2009001590, 14 October 2009) and conducted in accordance with the Declaration of Helsinki. Participants provided written consent after receiving a Participant Information and Consent Form.

      Data availability statement

      Access to the dataset is available by arrangement with the corresponding author. Owing to the identifying nature of the dataset, it will not be placed in an open-access repository.

      Conflict of Interest

      HPS is a shareholder of MoleMap NZ Limited and e-derm consult GmbH and undertakes regular teledermatological reporting for both companies. HPS is a Medical Consultant for Canfield Scientific Inc., a Medical Advisor for First Derm, and has a Medical Advisory Board Appointment with MoleMap NZ Limited. DLD, KJ, KJL, SRM, EKM, BD, AP, JER, DCW, MAB, NGM, SBS, HS, and RAS state no conflicts of interest.

      Acknowledgments

      We thank all the BNMS volunteers for their participation. This work was funded by NHMRC project grant numbers 1004999 , 1062935 , and the Centre of Research Excellence for the Study of Nevi 1099021 . The QSkin Study is supported by NHMRC program grant 1073898. The BLTS was supported most recently by NHMRC project 1031119 to NM. AP received a UQI tuition fee scholarship and a student living stipend. RAS was a Senior Research Fellow of the Australian NHMRC 1043187. HPS holds an NHMRC MRFF Next Generation Clinical Researchers Program Practitioner Fellowship 1137127. DCW is an NHMRC Principal Research Fellow 1058522, and MAB is an NHMRC Senior Principal Research Fellow 1024879. This research was carried out at the Translational Research Institute, Woolloongabba, Qld 4102, Australia. The Translational Research Institute is supported by a grant from the Australian Government .
      Funding bodies had no role in the study design, collection, analysis, and interpretation of the data, writing the report, or publication decisions.

      Author Contributions

      Conceptualization: HS, DLD, BMS, HPS, RAS; Data Curation:, DLD, KJ, KJL; Formal Analysis: DLD, KJ, RAS; Funding Acquisition: DLD, DCW, MAB, BMS, HS, HPS, RAS; Investigation: DLD, KJ, KJL, SRM, AP, JER; Methodology: DLD, DCW, HPS, RAS; Project Administration: HPS, RAS; Resources: EKM, BDA, DCW, MAB, NGM, BMS, HS, HPS, RAS; Software: DLD; Supervision: HPS, RAS; Validation: DLD, KJ, KJL, SRM, AP; Visualization: DLD; Writing - Original Draft Preparation: DLD, RAS; Writing - Review and Editing: DLD, KJ, KJL, SRM, EKM, BDA, AP, JER, DCW, MAB, NGM, BMS, HS, HPS, RAS

      Supplementary Material

      Introduction

      Epidemiological studies have shown that fair skin type, freckling, and high total body nevus counts (TNC) predispose to cutaneous melanoma (CM) (
      • Berwick M.
      • Buller D.B.
      • Cust A.
      • Gallagher R.
      • Lee T.K.
      • Meyskens F.
      • et al.
      Melanoma Epidemiology and Prevention.
      ). The progression from melanocyte to nevus cell toward CM can occur through several independent pathways (
      • Bennett D.C.
      Genetics of melanoma progression: the rise and fall of cell senescence.
      ;
      • Damsky W.E.
      • Bosenberg M.
      Melanocytic nevi and melanoma: unraveling a complex relationship.
      ;
      • Shain A.H.
      • Bastian B.C.
      From melanocytes to melanomas.
      ) that are each influenced by these host risk factors. The somatic genetic damage that drives CM development is strongly influenced by the density and type of skin melanin, pigmentation gene polymorphism, DNA damage response and propensity for nevus formation, and environmental sun exposure (
      • Shain A.H.
      • Joseph N.M.
      • Yu R.
      • Benhamida J.
      • Liu S.
      • Prow T.
      • et al.
      Genomic and transcriptomic analysis reveals incremental disruption of key signaling pathways during melanoma evolution.
      ;
      • Shain A.H.
      • Yeh I.
      • Kovalyshyn I.
      • Sriharan A.
      • Talevich E.
      • Gagnon A.
      • et al.
      The Genetic Evolution of Melanoma from Precursor Lesions.
      ;
      • Stark M.S.
      • Tan J.M.
      • Tom L.
      • Jagirdar K.
      • Lambie D.
      • Schaider H.
      • et al.
      Whole-exome sequencing of acquired nevi identifies mechanisms for development and maintenance of benign neoplasms.
      ). Although the level, intensity and frequency of skin exposure to solar UV radiation are key environmental risk factors for CM, there is not a linear dose relationship, with acute and/or intermittent sun exposure being disproportionately effective. The most important single genetic determinant of sun sensitivity and CM risk in populations of European ancestry is variation within the MC1R locus, where multiple alleles are associated with the red hair color phenotype (RHC).
      The morphological classification of the acquired melanocytic nevi observed through dermoscopy, including reticular, globular, and homogenous and/or complex (
      • Argenziano G.
      • Zalaudek I.
      • Ferrara G.
      • Hofmann-Wellenhof R.
      • Soyer H.P.
      Proposal of a new classification system for melanocytic naevi.
      ), has recently been the subject of an in-depth review (
      • Tan J.M.
      • Tom L.N.
      • Soyer H.P.
      • Stark M.S.
      Defining the Molecular Genetics of dermoscopic naevus patterns.
      ). These are the most common nevi that develop after birth, may be typical or atypical in appearance, and are considered benign. Histopathologically, these nevi are classed on the anatomical location within the skin, where they appear as discrete nests of melanocytic and/or nevus cells (
      • Damsky W.E.
      • Bosenberg M.
      Melanocytic nevi and melanoma: unraveling a complex relationship.
      ). The three major categories are junctional, where pigmented cells are confined to the epidermis; intradermal, where the cells are confined to the dermis; and compound, with both an epidermal and dermal component. Despite the histopathological differences, these acquired nevi have been thought to share a similar developmental pathway and relationship to melanoma, but independent pathways have been suggested to explain their formation (
      • Zalaudek I.
      • Hofmann-Wellenhof R.
      • Kittler H.
      • Argenziano G.
      • Ferrara G.
      • Petrillo L.
      • et al.
      A dual concept of nevogenesis: theoretical considerations based on dermoscopic features of melanocytic nevi.
      ). Two cytological features of nevus cells are clustering into nests and maturation. These features can provide a morphological basis to the patterns observed by dermoscopy (
      ). Maturation is a feature whereby there is a gradual change in cell shape and pigment content, with deeper lesions having a reduced nest size, cell, and nuclear content resulting in changes in cell shape (
      • Damsky W.E.
      • Bosenberg M.
      Melanocytic nevi and melanoma: unraveling a complex relationship.
      ). Longitudinal data have observed that reticular and globular nevi can change to the homogenous and/or structureless pattern with age (
      • Zalaudek I.
      • Catricalà C.
      • Moscarella E.
      • Argenziano G.
      What dermoscopy tells us about nevogenesis.
      ,
      • Zalaudek I.
      • Schmid K.
      • Marghoob A.A.
      • Scope A.
      • Manzo M.
      • Moscarella E.
      • et al.
      Frequency of dermoscopic nevus subtypes by age and body site: a cross-sectional study.
      ) and are likely to represent a heterogenous group of nevi. Globular nevi are typically compound or dermal nevi with the globules corresponding to the nest of melanocytic cells at the dermal-epidermal junction, papillary dermis, or entirely in the dermis. Reticular nevi are usually junctional with the pigment network corresponding to a layer of pigmented nevus and keratinocytes along the basal layer of the regularly elongated rete ridges of the dermal-epidermal junction, or nests of melanocytic cells accumulated at the tips of these elongated rete ridges, appearing as pigmented lines and the suprapapillary plates as less-pigmented holes, conferring the network appearance.
      It remains unclear how these different nevus types relate to other CM risk factors. It has been proposed that analysis of the nevus and dermoscopic diversity of each patient (intra-individual comparative analysis) would allow one to identify “the ugly duckling”, that is, suspicious lesions worthy of further attention (
      • Wazaefi Y.
      • Gaudy-Marqueste C.
      • Avril M.F.
      • Malvehy J.
      • Pellacani G.
      • Thomas L.
      • et al.
      Evidence of a limited intra-individual diversity of nevi: intuitive perception of dominant clusters is a crucial step in the analysis of nevi by dermatologists.
      ). The diameter of a nevus also has prognostic significance, with malignant transformation more likely in those larger than 5 mm (
      • Hofmann-Wellenhof R.
      • Marghoob A.A.
      • Zalaudek I.
      Large Acquired Nevus or Dysplastic Nevus: What's in the Name of a Nevus?.
      ), possibly mediated via increased telomere length (
      • Bataille V.
      • Kato B.S.
      • Falchi M.
      • Gardner J.
      • Kimura M.
      • Lens M.
      • et al.
      Nevus size and number are associated with telomere length and represent potential markers of a decreased senescence in vivo.
      ) or an inherent ability to overcome senescence (
      • Bonet C.
      • Luciani F.
      • Ottavi J.F.
      • Leclerc J.
      • Jouenne F.M.
      • Boncompagni M.
      • et al.
      Deciphering the Role of Oncogenic MITFE318K in Senescence Delay and Melanoma Progression.
      ). Using candidate analysis of MC1R variant alleles and 85 other genes (
      • Orlow I.
      • Satagopan J.M.
      • Berwick M.
      • Enriquez H.L.
      • White K.A.
      • Cheung K.
      • et al.
      Genetic factors associated with naevus count and dermoscopic patterns: preliminary results from the Study of Nevi in Children (SONIC).
      ), variants in several loci were associated with dermoscopic patterns observed during childhood, supporting the idea that genetic factors will strongly influence nevus morphology, as well as numbers in adults (
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      ).
      Polymorphisms within four genes, IRF4, MTAP, PLA2G6, and MITF, have been shown to strongly influence TNC. A single polymorphism within an intronic region of the IRF4 gene, rs12203592*C/T, shows a strong genotype-by-age interaction on the nevus count (
      • Duffy D.L.
      • Iles M.M.
      • Glass D.
      • Zhu G.
      • Barrett J.H.
      • Höiom V.
      • et al.
      IRF4 variants have age-specific effects on nevus count and predispose to melanoma.
      ) and is a major influence on pigmentation traits (
      • Praetorius C.
      • Grill C.
      • Stacey S.N.
      • Metcalf A.M.
      • Gorkin D.U.
      • Robinson K.C.
      • et al.
      A polymorphism in IRF4 affects human pigmentation through a tyrosinase-dependent MITF/TFAP2A pathway.
      ), tryosinase expression, and INFγ-induced gene regulation (
      • Chhabra Y.
      • Yong H.X.L.
      • Fane M.E.
      • Soogrim A.
      • Lim W.
      • Mahiuddin D.N.
      • et al.
      Genetic variation in IRF4 expression modulates growth characteristics, tyrosinase expression and interferon-gamma response in melanocytic cells.
      ,
      • Praetorius C.
      • Grill C.
      • Stacey S.N.
      • Metcalf A.M.
      • Gorkin D.U.
      • Robinson K.C.
      • et al.
      A polymorphism in IRF4 affects human pigmentation through a tyrosinase-dependent MITF/TFAP2A pathway.
      ). The association between TNC and single nucleotide polymorphism (SNPs) upstream of the MTAP gene, adjacent to the well-known tumor suppressor gene CDKN2A on chromosome 9p21, and the PLA2G6 gene on chromosome 22q13 exhibit less variable effects (
      • Newton-Bishop J.A.
      • Chang Y.M.
      • Iles M.M.
      • Taylor J.C.
      • Bakker B.
      • Chan M.
      • et al.
      Melanocytic nevi, nevus genes, and melanoma risk in a large case-control study in the United Kingdom.
      ). The fourth gene, MITF, has a coding region E318K allele that has also been shown to have a significant effect on nevus count, pigmentation, and melanomagenesis (
      • Bassoli S.
      • Pellegrini C.
      • Longo C.
      • Di Nardo L.
      • Farnetani F.
      • Cesinaro A.M.
      • et al.
      Clinical, dermoscopic, and confocal features of nevi and melanomas in a multiple primary melanoma patient with the MITF p.E318K homozygous mutation.
      ; Bonet et al., 2017;
      • Sturm R.A.
      • Fox C.
      • McClenahan P.
      • Jagirdar K.
      • Ibarrola-Villava M.
      • Banan P.
      • et al.
      Phenotypic characterization of nevus and tumor patterns in MITF E318K mutation carrier melanoma patients.
      ), but because the variant is uncommon, it has a lesser population level impact. Other nevus-associated genes have been recognized from a meta-analysis of 11 nevus genome-wide association study (GWAS) from Australia, Netherlands, United Kingdom, and the USA comprising over 50,000 phenotyped individuals (
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      ). This confirms that the strongest TNC-associated SNPs are within the IRF4, MTAP, and PLA2G6 genes and has detected new loci at or near a genome-wide level of significance.

      Materials and Methods

      Study samples

      Melanoma patients observed between October 2009 and November 2016 were recruited from the Melanoma Unit and Dermatology Department of the Princess Alexandra Hospital , private practices drawing clients from Southeast Queensland or the Queensland Cancer Registry (
      • Daley G.M.
      • Duffy D.L.
      • Pflugfelder A.
      • Jagirdar K.
      • Lee K.J.
      • Yong X.L.H.
      • et al.
      GSTP1 does not modify MC1R effects on melanoma risk.
      ,
      • Sturm R.A.
      • Fox C.
      • McClenahan P.
      • Jagirdar K.
      • Ibarrola-Villava M.
      • Banan P.
      • et al.
      Phenotypic characterization of nevus and tumor patterns in MITF E318K mutation carrier melanoma patients.
      ) (Supplementary Table S6). Controls (no personal but possible first degree family history of CM) were obtained by the advertisement of our study within the Princess Alexandra Hospital and by direct contact by letter of participants in the Brisbane Longitudinal Twin Study (
      • Duffy D.L.
      • Iles M.M.
      • Glass D.
      • Zhu G.
      • Barrett J.H.
      • Höiom V.
      • et al.
      IRF4 variants have age-specific effects on nevus count and predispose to melanoma.
      ) or QSkin participants (
      • Olsen C.M.
      • Green A.C.
      • Neale R.E.
      • Webb P.M.
      • Cicero R.A.
      • Jackman L.M.
      • et al.
      Cohort profile: the QSkin Sun and Health Study International.
      ).

      Measures and dermoscopy

      All participants underwent detailed physical examination, including whole body digital images, nevus counts, and dermoscopic imaging performed by a trained research assistant on nevi ≥ 5mm in diameter. Photographs from 16 body sites and individual dermoscopic images of all significant nevi were recorded for 793 participants with a system for sequential total body photography and dermoscopy (FotoFinder Systems GmbH, Bad Birnbach, Germany) and a 3-dimensional total body image and dermoscopic images for 471 participants using a Vectra WB360 3-dimensional body imager (Canfield Scientific , Parsippany, NJ). As described in detail elsewhere (
      • McWhirter S.R.
      • Duffy D.L.
      • Lee K.J.
      • Wimberley G.
      • McClenahan P.
      • Ling N.
      • et al.
      Classifying dermoscopic patterns of naevi in a case-control study of melanoma.
      ), all the nevi ≥ 5mm were classified in terms of the predominant dermoscopic pattern using one of two methods (old and new BNMS [Brisbane Nevus Morphology Study] classification systems) that were integrated for analysis. Counting all nevi ≥ 2mm on the back in a subset of 572 individuals showed a high correlation with TNC ≥ 5mm (Supplementary Table S7). A clinical assessment using standardized pigmentation characteristics was performed as described (
      • Ainger S.A.
      • Jagirdar K.
      • Lee K.J.
      • Soyer H.P.
      • Sturm R.A.
      Skin pigmentation genetics for the clinic.
      ,
      • Koh U.
      • Janda M.
      • Aitken J.F.
      • Duffy D.L.
      • Menzies S.
      • Sturm R.A.
      • et al.
      Mind your Moles' study: protocol of a prospective cohort study of melanocytic nevi.
      ,
      • Sturm R.A.
      • Fox C.
      • McClenahan P.
      • Jagirdar K.
      • Ibarrola-Villava M.
      • Banan P.
      • et al.
      Phenotypic characterization of nevus and tumor patterns in MITF E318K mutation carrier melanoma patients.
      ).

      Biospecimen collection

      We report the analysis of 1,266 participants representing the BNMS (including 27 monozygotic and 39 dizygotic twins), 1,235 of whom provided saliva samples using an Oragene-DNA self-collection kit (DNA Genotec, Ottowa, Canada) sufficient for genomic DNA processing. Genomic DNA was extracted from 2 ml of saliva collected, and its quality and quantity was checked using Nanodrop UV absorbance and a Qubit fluorometer (Thermo Fisher, Waltham, MA). A minimum of 2.5 ug of DNA was then used for genotyping on the Infinium HumanCoreExome-24 Microarray (Illumina, San Diego, CA).

      DNA sequencing and genotyping

      Large scale SNP genotyping was performed using the Infinium Microarray HumanCoreExome-24 Chip (Illumina) as previously described (
      • Daley G.M.
      • Duffy D.L.
      • Pflugfelder A.
      • Jagirdar K.
      • Lee K.J.
      • Yong X.L.H.
      • et al.
      GSTP1 does not modify MC1R effects on melanoma risk.
      ), and the data was manipulated using the PLINK program (
      • Chang C.C.
      • Chow C.C.
      • Tellier L.C.
      • Vattikuti S.
      • Purcell S.M.
      • Lee J.J.
      Second-generation PLINK: rising to the challenge of larger and richer datasets.
      ). Genotype data were cleaned using a standard pipeline excluding SNPs exhibiting extreme Hardy-Weinberg disequilibrium and between-batch heterogeneity. Single SNP genotyping was performed using TaqMan SNP genotyping assays (Thermo Fisher Scientific, Waltham, MA) (
      • Cook A.L.
      • Chen W.
      • Thurber A.E.
      • Smit D.J.
      • Smith A.G.
      • Bladen T.G.
      • et al.
      Analysis of cultured human melanocytes based on polymorphisms within the SLC45A2/MATP, SLC24A5/NCKX5, and OCA2/P loci.
      ) in a 96 or 384 well plate format using a 7500 real-time PCR system and analyzed using 7500 Software (Applied Biosystems, Foster City, CA). The MC1R coding variants V60L, R163Q, and D294H, not included on the Illumina array, were genotyped by TaqMan assay C_7519054_10, C_7519060_20, and C_1010501_30 (ThermoFisher), respectively. Half the sample set were also subjected to automated DNA sequence analysis of PCR-amplified fragments covering the MC1R coding region (
      • Box N.F.
      • Duffy D.L.
      • Chen W.
      • Stark M.
      • Martin N.G.
      • Sturm R.A.
      • et al.
      MC1R genotype modifies risk of melanoma in families segregating CDKN2A mutations.
      ), and where identified MITF E318K variant rs149617956*G/A containing exon 9 (
      • Yokoyama S.
      • Woods S.L.
      • Boyle G.M.
      • Aoude L.G.
      • MacGregor S.
      • Zismann V.
      • et al.
      A novel recurrent mutation in MITF predisposes to familial and sporadic melanoma.
      ) using ABI PRISM Big Dye Terminator Sequencing (Applied Biosystems) with reactions processed by the Australian Equine Genetics Research Centre (AEGRC, Brisbane, Australia).
      For the melanoma case-control analysis, we used an external set of general population controls (i.e., unknown melanoma status) to maximize the statistical power (this was not possible for dermoscopically classified nevus counts). Australian samples were from the QIMR Brisbane Longitudinal Twin Study (
      • Daley G.M.
      • Duffy D.L.
      • Pflugfelder A.
      • Jagirdar K.
      • Lee K.J.
      • Yong X.L.H.
      • et al.
      GSTP1 does not modify MC1R effects on melanoma risk.
      ) and QSkin samples (
      • Olsen C.M.
      • Pandeya N.
      • Thompson B.S.
      • Dusingize J.C.
      • Green A.C.
      • Neale R.E.
      • et al.
      Association between phenotypic Characteristics and Melanoma in a Large Prospective Cohort Study.
      ). In the case of the MITF rs149617956 SNP, instead we used the United Kingdom Biobank frequency (
      • Bycroft C.
      • Freeman C.
      • Petkova D.
      • Band G.
      • Elliott L.T.
      • Sharp K.
      • et al.
      The UK Biobank resource with deep phenotyping and genomic data.
      ), as this was not available on most of the external genotyping arrays. We excluded all SNPs exhibiting Hardy-Weinberg disequilibrium in controls with exact Hardy-Weinberg test P < 0.0001 and/or absent homozygotes despite the presence of heterozygotes, and those individuals with >3% genotyping failure. We retained related individuals, and the analyses have appropriately adjusted for this fact, such as a GEMMA mixed model analysis using the empirical kinship matrices and generalized linear mixed models or gene-dropping based on known pedigree relationships. The presented GWAS only used directly genotyped markers, but in addition, we have imputed key candidate SNPs from the regions reported to be associated with melanoma or total nevus count in Duffy et al (2018). The imputation of these regions was performed using BEAGLE5 (
      • Browning B.L.
      • Zhou Y.
      • Browning S.R.
      A one-penny imputed genome from next-generation reference panels.
      ).

      Statistical analysis of datasets

      Statistical analyses were carried out in the R Statistical Computing Environment (version 2.6.0, R Foundation for Statistical Computing, Vienna, Austria) and the GEMMA mixed model association computer package (
      • Zhou X.
      • Stephens M.
      Genome-wide efficient mixed-model analysis for association studies.
      ,
      • Zhou X.
      • Stephens M.
      Efficient multivariate linear mixed model algorithms for genome-wide association studies.
      ). Because we include related cases and controls in our analyses, linear and logistic regression analyses were carried out using the lme4 package in R (
      • Bates D.
      • Mächler M.
      • Bolker B.
      • Walker S.
      Fitting linear mixed-effects models using lme4.
      ). Based on the results from Box-Cox regression analyses, we chose to log transform the nevus counts (log10[count+1]). Since the nevus counts varied by age in a nonlinear fashion, we included linear and quadratic age terms as covariates, along with sex and the nevus counting protocol. For the genome-wide association analyses, we included melanoma case status as a covariate, since this variable has such a strong association with the nevus count, as well as the first four principal components. To perform the SNP genotype-based principal components analysis of ancestry, we used the TRACE computer package (
      • Wang C.
      • Zhan X.
      • Liang L.
      • Abecasis G.R.
      • Lin X.
      Improved ancestry estimation for both genotyping and sequencing data using projection procrustes analysis and genotype imputation.
      ). This estimates principal components scores for our study participants with respect to an external reference panel (the Human Genome Diversity Panel). The default four principal component scores from this analysis were used as covariates in the association analyses. Alhough the mean age is different between the case and control, there are sufficient numbers in each age band to allow us to correctly model the relationship between age and nevus count, as discussed elsewhere (
      • Duffy D.L.
      • Lee K.J.
      • Jagirdar K.
      • Pflugfelder A.
      • Stark M.S.
      • McMeniman E.K.
      • et al.
      High naevus count and MC1R red hair alleles contribute synergistically to increased melanoma risk.
      ). Aside from TNC, the variables that we study are constant over age, such as genotype. Q-Q plots for the GWAS analysis are presented in Supplementary Figure S3. A P < 5x10-8 for each SNP was used as the cut off for genome-wide statistical significance, and P < 5x10-7 suggestive of significance. For 33 candidate SNPs, the Bonferroni corrected critical threshold for one trait (experiment-wise α < 0.05) would be 0.0015.

      Data availability statement

      Access to the dataset is available by arrangement with the corresponding author. Owing to the identifying nature of the dataset, it will not be placed in an open-access repository.

      Results

      BNMS participants’ demographics and phenotype

      The study sample consisted of 1,266 individuals with counts of nevi of different dermoscopic morphology, the majority of European ancestry with minor contributions from Asia and Oceania (Supplementary Figure S4). There were 607 CM cases and 659 unaffected controls (Supplementary Table S1), with an enrichment of multiple primary melanoma cases; 275 participants (45% of the case participants) had two or more CM. The average age of the cases (55.5 years) was older than the controls (39.7 years), reflecting the oversampling (Supplementary Figure S5) of younger controls from the Brisbane Longitudinal Twin Study (
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      ) for whom we have longitudinal TNC (
      • Lee S.
      • Duffy D.L.
      • McClenahan P.
      • Lee K.J.
      • McEniery E.
      • Burke B.
      • et al.
      Heritability of naevus patterns in an adult twin cohort from the Brisbane Twin Registry: a cross-sectional study.
      ). There were slightly more male cases (52.7%) but fewer male controls (44.6%). Skin (fair), hair (red), and eye (blue/gray) color were all associated with CM risk, with the frequency distribution of hair color in the control participants closely matching that from other Australian population studies (Supplementary Table S8). Melanoma cases had more than twice as many nevi ≥ 5mm diameter at any age (average case 31.4, control 13.3) and more freckles. The cases were heavier than the controls (average body mass index 28.1 vs 25.2), again reflecting the average age difference, but did not differ from the age-matched Queenslanders (
      • Olsen C.M.
      • Green A.C.
      • Neale R.E.
      • Webb P.M.
      • Cicero R.A.
      • Jackman L.M.
      • et al.
      Cohort profile: the QSkin Sun and Health Study International.
      ).

      Nevus subtypes called in this study and relationship to melanoma

      The distribution plots of TNC in the cases and controls by age and dermoscopic nevus subtype are shown in Supplementary Figure S1. In each case for TNC, nonspecific, and reticular subtype regression curves for the cases are higher than those of the controls, with counts peaking at approximately age 50 years and then declining. The exception to this pattern was the globular subtype, in which the regression curves for both the cases and controls decline from 20 years of age, consistent with other reports (
      • Zalaudek I.
      • Grinschgl S.
      • Argenziano G.
      • Marghoob A.A.
      • Blum A.
      • Richtig E.
      • et al.
      Age-related prevalence of dermoscopy patterns in acquired melanocytic naevi.
      ,
      • Zalaudek I.
      • Schmid K.
      • Marghoob A.A.
      • Scope A.
      • Manzo M.
      • Moscarella E.
      • et al.
      Frequency of dermoscopic nevus subtypes by age and body site: a cross-sectional study.
      ).

      GWAS of nevus subtypes in the BNMS and melanoma risk

      The two SNPs on chromosome 1, rs4970612 and rs7545703, reached suggestive significance for TNC and reticular but not nonspecific nevus types, indicating a possible reticular-specific effect. However, these singleton SNPs were not associated with TNC in the large nevus meta-analysis (
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      ) and do not lie in regions of functional significance for melanocytes (Supplementary Table S9).

      Effect of Nevogenic Genotype on TNC and Nevus Subtypes

      We have plotted the nevus count distribution by the genotype of SNPs from the three genes most highly linked with TNC or a nevus subtype (Supplementary Table S5), (i) MTAP rs7023329*A/G, (ii) PLA2G6 rs132985*C/T, and (iii) IRF4 rs12203592*C/T (Supplementary Figure S6, Supplementary Figure S7, Supplementary Table S10). The significant association with globular nevi for IRF4 is seen by rs12203592*C/C as having greater number and rs12203592*T/T as the lowest median globular counts (β = 9.55x10-2). MTAP rs7023329*A/A versus *G/G (β = 5.24x10-2) and PLA2G6 rs132985*C/C versus *T/T (β = 4.32x10-2) show similar but smaller effects on the median globular counts. Consistent with this, the MTAP rs7023329*A/A and PLA2G6 rs132985*C/C also had the greater number of TNC and reticular nevus counts. However, the rs12203592*T/T genotype was associated with the lowest globular nevi and the greatest number of nonspecific nevi.
      Although the MC1R peak SNP associated with CM was not in itself linked to TNC or any dermoscopic nevus subtype (Supplementary Table S5), we have previously shown that when considered by MC1R variant allele classification as MC1R*WT, MC1R*r, and MC1R*R (
      • Ainger S.A.
      • Jagirdar K.
      • Lee K.J.
      • Soyer H.P.
      • Sturm R.A.
      Skin pigmentation genetics for the clinic.
      ), the MC1R*R/r genotype does act to modify TNC (
      • Duffy D.L.
      • Lee K.J.
      • Jagirdar K.
      • Pflugfelder A.
      • Stark M.S.
      • McMeniman E.K.
      • et al.
      High naevus count and MC1R red hair alleles contribute synergistically to increased melanoma risk.
      ). The MC1R*R/r and MC1R*R/R genotypes, which have the highest association with CM, were demonstrated to have a significant association with the nonspecific nevus subtype compared to the MC1R*WT genotype (P = 1.4x10-8) using the model in Supplementary Table S11.

      Interaction of nevus genes in TNC

      The genetic interaction between MTAP and PLA2G6 TNC-associated SNPs in the determination of nevus counts can be observed in Supplementary Figure S8, which plots the median TNC by combined genotype. There is a progressive decrease in the count from a maximum of 25 observed with MTAP rs7023329*A/A, PLA2G6 rs132985*C/C risk genotype to a minimum of 8 observed with MTAP rs7023329*G/G, PLA2G6 rs132985*T/T, with the stepwise progression the most obvious with the heterozygosity of MTAP rs7023329*A/G with the changing genotype at PLA2G6 rs132985*C/C, *C/T, *T/T (green shading in Supplementary Figure S8).

      Discussion

      The observation that host factors, including pigmentary skin type, are associated with the nevus count, density, distribution, and dermoscopic subtype (
      • Zalaudek I.
      • Argenziano G.
      • Mordente I.
      • Moscarella E.
      • Corona R.
      • Sera F.
      • et al.
      Nevus type in dermoscopy is related to skin type in white persons.
      ,
      • Zalaudek I.
      • Catricalà C.
      • Moscarella E.
      • Argenziano G.
      What dermoscopy tells us about nevogenesis.
      ,
      • Zalaudek I.
      • Schmid K.
      • Marghoob A.A.
      • Scope A.
      • Manzo M.
      • Moscarella E.
      • et al.
      Frequency of dermoscopic nevus subtypes by age and body site: a cross-sectional study.
      ) supports a role for genetic factors in the development and morphology of acquired melanocytic lesions. Given the effect sizes of known loci on TNC, we expected the statistical power of our study to be sufficient to detect the effects on subtype counts only in the candidate loci. Fortunately, we had results from our earlier large meta-analysis to guide us in selecting 33 chromosomal regions (including MITF): 13 of 33 (39%) gave a P-value < 0.01 for melanoma, 5 of 33 (15%) for TNC, 5 of 33 (15%) for nonspecific nevi, 4 of 33 (12%) for globular nevi, and 2 of 33 (6%) for reticular nevi. The MTAP gene reached significance for all nevus properties assayed, the PLA2G6 and MITF genes for three, and IRF4 and SLC45A2 for two. All the genes associated with the nevus phenotypes were also significantly associated with melanoma.
      Multiple GWAS have implicated the SNP rs12203592*C/T within the fourth intron of the IRF4 gene to be associated with skin pigmentation, hair and eye color, UVR sensitivity, CM susceptibility, tanning and facial pigmented spots (
      • Han J.
      • Kraft P.
      • Nan H.
      • Guo Q.
      • Chen C.
      • Qureshi A.
      • et al.
      A genome-wide association study identifies novel alleles associated with hair color and skin pigmentation.
      ,
      • Jacobs L.C.
      • Hamer M.A.
      • Gunn D.A.
      • Deelen J.
      • Lall J.S.
      • van Heemst D.
      • et al.
      A Genome-Wide Association study identifies the skin color genes IRF4, MC1R, ASIP, and BNC2 influencing facial pigmented spots.
      ,
      • Nan H.
      • Kraft P.
      • Qureshi A.A.
      • Guo Q.
      • Chen C.
      • Hankinson S.E.
      • et al.
      Genome-wide association study of tanning phenotype in a population of European ancestry.
      ), nevus counts, and iris freckling (
      • Duffy D.L.
      • Iles M.M.
      • Glass D.
      • Zhu G.
      • Barrett J.H.
      • Höiom V.
      • et al.
      IRF4 variants have age-specific effects on nevus count and predispose to melanoma.
      ,
      • Laino A.M.
      • Berry E.G.
      • Jagirdar K.
      • Lee K.J.
      • Duffy D.L.
      • Soyer H.P.
      • et al.
      Iris pigmented lesions as a marker of cutaneous melanoma risk: an Australian case-control study.
      ). In this study, we see that *T/T carriers have a lower TNC, especially in our CM cases (interaction P = 0.005, Supplementary Tables S12 and S13, Supplementary Figure S9). Recently, the rs12203592*T allele has been associated with CM, solar elastosis, and inversely with neval remnants (
      • Gibbs D.C.
      • Orlow I.
      • Bramson J.I.
      • Kanetsky P.A.
      • Luo L.
      • Kricker A.
      • et al.
      Association of interferon regulatory Factor-4 polymorphism rs12203592 with divergent melanoma pathways.
      ), increased Breslow thickness of tumors (
      • Gibbs D.C.
      • Ward S.V.
      • Orlow I.
      • Cadby G.
      • Kanetsky P.A.
      • Luo L.
      • et al.
      Functional melanoma-risk variant IRF4 rs12203592 associated with Breslow thickness: a pooled international study of primary melanomas.
      ), and decreased CM survival (
      • Potrony M.
      • Rebollo-Morell A.
      • Giménez-Xavier P.
      • Zimmer L.
      • Puig-Butille J.A.
      • Tell-Marti G.
      • et al.
      IRF4 rs12203592 functional variant and melanoma survival.
      ). In melanocytic cells, we previously assigned a function for the IRF4 protein as a transcriptional regulator of the TYR gene (
      • Chhabra Y.
      • Yong H.X.L.
      • Fane M.E.
      • Soogrim A.
      • Lim W.
      • Mahiuddin D.N.
      • et al.
      Genetic variation in IRF4 expression modulates growth characteristics, tyrosinase expression and interferon-gamma response in melanocytic cells.
      ,
      • Praetorius C.
      • Grill C.
      • Stacey S.N.
      • Metcalf A.M.
      • Gorkin D.U.
      • Robinson K.C.
      • et al.
      A polymorphism in IRF4 affects human pigmentation through a tyrosinase-dependent MITF/TFAP2A pathway.
      ), with the rs12203592*T allele reducing the IRF4 protein expression. There were also effects of rs12203592*T on melanocyte strain growth, response to UVR treatment, and IFNγ cytokine induction. It is highly likely that there is a melanocyte intrinsic effect of the rs12203592*T allele on globular nevus formation and involution. IRF4 is best known for its immunological roles, and IFNγ is likely to affect melanocyte behavior (
      • Natarajan V.T.
      • Ganju P.
      • Singh A.
      • Vijayan V.
      • Kirty K.
      • Yadav S.
      • et al.
      IFN-gamma signaling maintains skin pigmentation homeostasis through regulation of melanosome maturation.
      ,
      • Son J.
      • Kim M.
      • Jou I.
      • Park K.C.
      • Kang H.Y.
      IFN-gamma inhibits basal and alpha-MSH-induced melanogenesis.
      ). We demonstrated that elicited immune cell responses in the skin are modulated by the rs12203592 genotype (
      • Chhabra Y.
      • Yong H.X.L.
      • Fane M.E.
      • Soogrim A.
      • Lim W.
      • Mahiuddin D.N.
      • et al.
      Genetic variation in IRF4 expression modulates growth characteristics, tyrosinase expression and interferon-gamma response in melanocytic cells.
      ), but the genotypic effects of IRF4 on the melanocytic DNA repair (cf MC1R’s known role) are yet to be studied. Notably, the characterization of somatic genetic damage has shown significantly fewer copy number aberrations in globular compared to reticular nevi (
      • Stark M.S.
      • Tan J.M.
      • Tom L.
      • Jagirdar K.
      • Lambie D.
      • Schaider H.
      • et al.
      Whole-exome sequencing of acquired nevi identifies mechanisms for development and maintenance of benign neoplasms.
      ), implicating an IRF4 polymorphism in this process.
      The MTAP rs7023329 and PLA2G6 rs11570734 SNPs were found to be the most informative for predicting TNC in the BNMS, with the MTAP gene polymorphism associated with all the dermoscopic nevus patterns. The function of these genes in the formation of nevi and CM is not known, but in the case of MTAP, these are linked to differential gene expression levels driven by elements linked to rs7023329 (
      • Sangalli A.
      • Malerba G.
      • Tessari G.
      • Rodolfo M.
      • Gomez-Lira M.
      Melanoma risk alleles are associated with downregulation of the MTAP gene and hypermethylation of a CpG island upstream of the gene in dermal fibroblasts.
      ). PLA2G6 encodes iPLA2β-VIA, a protein that is essential for membrane remodeling in neurons and that has variable enzymatic activity depending on which phospholipids comprise these membranes. Morphologically characteristic axonal degeneration in a group of “PLA2G6-associated neurodegeneration” (PLAN) diseases is thought to be because of the degeneration of specific membranes (
      • Sumi-Akamaru H.
      • Beck G.
      • Kato S.
      • Mochizuki H.
      Neuroaxonal dystrophy in PLA2G6 knockout mice.
      ).
      Our earlier findings supported the hypothesis that that individuals carrying the MC1R RHC genotypes tend to have fewer nevi (
      • Duffy D.L.
      • Box N.F.
      • Chen W.
      • Palmer J.S.
      • Montgomery G.W.
      • James M.R.
      • et al.
      Interactive effects of MC1R and OCA2 on melanoma risk phenotypes.
      ). In this study, we found that the RHC allele homozygotes (MC1R*R/R) had lower counts than the MC1R*R heterozygote carriers, but this latter group had higher counts than the wild type carriers on average (Supplementary Table S11). This seems consistent with a mixture of mechanisms acting on skin color lightening and longitudinal effects on nevus counts, as has been suggested in the case of IRF4 genotype and CM risk (
      • Duffy D.L.
      • Iles M.M.
      • Glass D.
      • Zhu G.
      • Barrett J.H.
      • Höiom V.
      • et al.
      IRF4 variants have age-specific effects on nevus count and predispose to melanoma.
      ,
      • Gibbs D.C.
      • Ward S.V.
      • Orlow I.
      • Cadby G.
      • Kanetsky P.A.
      • Luo L.
      • et al.
      Functional melanoma-risk variant IRF4 rs12203592 associated with Breslow thickness: a pooled international study of primary melanomas.
      ). In another study of nevus and freckle phenotypes in children (
      • Barón A.E.
      • Asdigian N.L.
      • Gonzalez V.
      • Aalborg J.
      • Terzian T.
      • Stiegmann R.A.
      • et al.
      Interactions between ultraviolet light and MC1R and OCA2 variants are determinants of childhood nevus and freckle phenotypes.
      ), it was reported that sun exposure patterns can interact with MC1R RHC and OCA2 variant genotypes to influence TNC, with the report that MC1R*R/r, MC1R*R/R together with homozygosity for OCA2 rs12913832*C blue eye color alleles had elevated nevus counts. The rs4778138 SNP assayed here as the most informative SNP within the OCA2 locus was significant at P < 0.05 for CM, TNC, and nonspecific nevi (Supplementary Table S5) in adults. However, the MC1R*R/r, MC1R*R/R genotypes were a greater influence, increasing nonspecific nevus counts in the BNMS (Supplementary Table S11). MC1R RHC variants have been shown to be associated with hypopigmentation, reduced structure of dermoscopic patterns, and more discernible vessels in benign, atypical, and CM lesions (
      • Bassoli S.
      • Maurichi A.
      • Rodolfo M.
      • Casari A.
      • Frigerio S.
      • Pupelli G.
      • et al.
      CDKN2A and MC1R variants influence dermoscopic and confocal features of benign melanocytic lesions in multiple melanoma patients.
      ,
      • Cuéllar F.
      • Puig S.
      • Kolm I.
      • Puig-Butille J.
      • Zaballos P.
      • Martí-Laborda R.
      • et al.
      Dermoscopic features of melanomas associated with MC1R variants in Spanish CDKN2A mutation carriers.
      ,
      • Quint K.D.
      • Van der Rhee J.I.
      • Gruis N.A.
      • Ter Huurne J.A.
      • Wolterbeek R.
      • Van der Stoep N.
      • et al.
      Melanocortin 1 receptor (MC1R) variants in high melanoma risk patients are associated with specific dermoscopic ABCD features.
      ). Another recent study of the influence of the MC1R genotype on the dermoscopic features of nevi found that R-allele carriers had visible vessels, dots, and globules present in their lesions, with eccentric hyperpigmentation (
      • Vallone M.G.
      • Tell-Marti G.
      • Potrony M.
      • Rebollo-Morell A.
      • Badenas C.
      • Puig-Butille J.A.
      • Gimenez-Xavier P.
      • et al.
      Melanocortin 1 receptor (MC1R) polymorphisms' influence on size and dermoscopic features of nevi.
      ). Together, these studies and the BNMS (
      • Duffy D.L.
      • Lee K.J.
      • Jagirdar K.
      • Pflugfelder A.
      • Stark M.S.
      • McMeniman E.K.
      • et al.
      High naevus count and MC1R red hair alleles contribute synergistically to increased melanoma risk.
      ) indicate that MC1R RHC alleles are correlated with the nevus phenotype, including TNC, dermoscopic pattern, and body site distribution (
      • Vallone M.G.
      • Tell-Marti G.
      • Potrony M.
      • Rebollo-Morell A.
      • Badenas C.
      • Puig-Butille J.A.
      • Gimenez-Xavier P.
      • et al.
      Melanocortin 1 receptor (MC1R) polymorphisms' influence on size and dermoscopic features of nevi.
      ).
      Overall, the range of genes and biological pathways identified in this study of TNC and nevus morphology indicate that multiple biological pathways will be involved in nevogenesis (
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      ).
      Figure thumbnail fx1
      Supplementary Figure S1Total nevus count by dermoscopic subtype and age. Scatter plot of TNC and dermoscopic nevus subtype body counts by age. Nevus count expressed on a logarithmic scale on the y-axis and age in years on the x-axis. Locally weighted regression curves for counts versus age in CM (full line) or control groups (dashed line), with colors representing nevus subtypes as indicated in the legend. CM, cutaneous melanoma.
      Figure thumbnail fx2
      Supplementary Figure S2GWAS plot for reticular and nonspecific nevus count. Manhattan plot of the P-values for (a) reticular and (b) nonspecific nevus count ≥ 5mm with genome-wide significance indicated by the black dashed line (5x10-8) and suggestive of significance (5x10-7) by the gray dotted line. Blue dots indicate the SNPs within previously association peaks, and the gray dots represent all the other SNPs on the Illumina HumanCoreExome array. The peak-associated SNP rs7023329 on chromosome 9 is indicated as MTAP. GWAS, genome-wide association study; SNP, single nucleotide polymorphism.
      Figure thumbnail fx3
      Supplementary Figure S3Q-Q plots of GWAS for total and subtype nevus counts. The quantile-quantile (Q-Q) plots for genome-wide mixed model association analysis X-square of the four phenotypes: log-transformed TNC, nonspecific, globular, and reticular nevus counts. GWAS, genome-wide association study; TNC, total body nevus count.
      Figure thumbnail fx4
      Supplementary Figure S4Ancestry analysis by geneticprincipal components analysis. Plot of the first two components from the genetic principal components analysis of the BNMS participants CM cases (red) and unaffected (blue). These were supplemented by an additional 938 samples from the human genome diversity project in the TRACE computer package (gray) to maximize power to estimate the scores for BNMS participants. BNMS, Brisbane Nevus Morphology Study; CM, cutaneous melanoma; PC, principal component.
      Figure thumbnail fx5
      Supplementary Figure S5Number of cases and controls at each age and its effect on the estimation of the relationship between age and nevus count ≥ 5mm. The upper panel provides counts of participation by age. The lower panel is a scatter plot of TNC ≥ 5mm by age and provides local confidence bands for age-nevus localized regression analysis fitted separately to cases and to controls using the R locfit package. The curves are roughly parallel over the entire age range, and shaded confidence bands of similar width except past the age of 70. For the critical phenotype of the nevus count, there is no evidence for bias or interaction that would affect the analysis combining cases and controls. TNC, total body nevus count.
      Figure thumbnail fx6
      Supplementary Figure S6MTAP, PLA2G6, and IRF4 association with nevus dermoscopic subtypes. Three SNPs in genes previously linked with TNC were used to test for the association with the nevus count ≥ 5mm in the BNMS cohort. The box-and-whisker plots show the age, sex and classification method (
      • McWhirter S.R.
      • Duffy D.L.
      • Lee K.J.
      • Wimberley G.
      • McClenahan P.
      • Ling N.
      • et al.
      Classifying dermoscopic patterns of naevi in a case-control study of melanoma.
      ) adjusted median TNC on a logarithmic scale (solid bar; y-axis) presented by genotype (x-axis) for (a) MTAP rs7023329*A/G, (b) PLA2G6 rs132985*C/T, (c) IRF4 rs12203592*C/T in the samples subdivided into globular, reticular and nonspecific. The 75% to 25% limits are indicated by the upper and lower boxes and an approximate 95% confidence interval for the median by the vertical range of notches in the sides of the boxes around the median. The 97.5% upper and 2.5% lower range are the upper and lower extents of the vertical lines (whiskers) outside of the boxes with the outliers represented by empty circles. BNMS, Brisbane Nevus Morphology Study; SNP, single nucleotide polymorphism; TNC, total body nevus count.
      Figure thumbnail fx7
      Supplementary Figure S7MTAP, PLA2G6, and IRF4 association with the total nevus count. The three SNPs from Figure 4 were used to test for association with TNC ≥ 5mm in the BNMS cohort. The box-and-whisker plots show the age, sex, and classification method (
      • McWhirter S.R.
      • Duffy D.L.
      • Lee K.J.
      • Wimberley G.
      • McClenahan P.
      • Ling N.
      • et al.
      Classifying dermoscopic patterns of naevi in a case-control study of melanoma.
      ) adjusted median TNC on a logarithmic scale (solid bar; y-axis) presented by genotype (x-axis) for (a) MTAP rs7023329*A/G, (b) PLA2G6 rs132985*C/T, and (c) IRF4 rs12203592*C/T. BNMS, Brisbane Nevus Morphology Study; SNP, single nucleotide polymorphism; TNC, total body nevus count.
      Figure thumbnail fx8
      Supplementary Figure S8MTAP and PLA2G6 genotype have a synergistic effect on the nevus count. The combined genotypes at MTAP rs7023329*A/G and PLA2G6 rs132985*C/T show a synergistic effect on the nevus count ≥5mm in the BNMS cohort. The box-and-whisker plots show the age, sex, and classification method (
      • McWhirter S.R.
      • Duffy D.L.
      • Lee K.J.
      • Wimberley G.
      • McClenahan P.
      • Ling N.
      • et al.
      Classifying dermoscopic patterns of naevi in a case-control study of melanoma.
      ) adjusted median TNC on a logarithmic scale (solid bar; y-axis) presented by genotype (x-axis). The MTAP rs7023329*A/A genotype is shaded blue, the rs7023329*A/T genotype shaded green, and the rs7023329*T/T genotype shaded yellow. BNMS, Brisbane Nevus Morphology Study; TNC, total body nevus count.
      Figure thumbnail fx9
      Supplementary Figure S9IRF4 association with the total nevus count in melanoma case participants. The association of IRF4 rs12203592*C/T genotype with TNC is shown in the box-and-whiskers plot for cases only and unadjusted for any covariates. TNC, total body nevus count.
      Supplementary Table S1Demographic and Phenotypic Characteristics of 1,266 BNMS Participants
      CharacteristicTotal Mean (SD) or N (%)Unaffected
      Unaffected, participant with no prior history of melanoma
      Mean (SD) or N (%)
      Case
      Case, participant with prior history of at least one melanoma
      Mean (SD) or N (%)
      P-value
      Unadjusted for other covariates
      Age (years)< 10-6
       Male49·9 ± 17·839·7 ± 15·059·2 ± 14·8
       Female44·8 ± 15·639·6 ± 14·351·5 ± 14·6
       Total47·3 ± 16·939·7 ± 14·655·5 ± 15·2
       Range11-8811-7614-88
      Total N1266659607
      Sex0·007
       Male607 (48·4)294 (44·6)320 (52·7)
       Female659 (51·6)365 (55·7)287 (47·3)
      Total N1266659607
      Skin color0·015
       Fair977 (77·7)497 (75·5)480 (80·1)
       Medium242 (19·3)133 (20·2)109 (18·2)
       Olive38 (3·0)28 (4·3)10 (1·7)
      Total N1257658599
      Hair color1x10-5
       Red146 (11·5)47 (7·2)99 (16·4)
       Blonde236 (18·8)119 (18·1)117 (19·4)
       Light brown418 (33·0)240 (36·6)178 (29·5)
       Dark brown400 (31·6)219 (33·4)181 (30·0)
       Black59 (4·7)31 (4·7)28 (4·6)
      Total N1259656603
      Eye color7x10-5
       Blue/gray676 (53·4)329 (50·0)347 (57·2)
       Green/hazel371 (29·3)187 (28·4)184 (30·4)
       Brown217 (17·1)142 (21·6)75 (12.4)
      Total N1264658606
      Total Nevi Count ≥5mm mean +/- SD< 10-6
       Globular2.3 +/- 5.21.8 +/- 4.12.7 +/- 6.2
       Reticular3.6 +/- 8.72.2 +/- 5.75.2 +/- 10.9
       Non-specific16.4 +/- 22.49.3 +/- 13.624.2 +/- 27.1
       Total22.0 +/- 28.513.3 +/- 18.931.4 +/- 33.7
      Total N1266659607
      Combined Freckling Score
      Face, Shoulder, Hands (none = 0, mild = 1, moderate = 2, severe =3; 0 to 9 total)
      < 10-6
       0 to 3726 (59·1)432 (66·1)294 (51·1)
       3 to 6406 (33·0)178 (27·2)228 (39·7)
       7 to 997 (7·9)44 (6·7)53 (9·2)
      Total N1229654575
      Height (cm)0·02
       Male178·0 ± 7·1179·6 ± 6·9176·6 ± 6·8
       Female165·7 ± 6·9166·3 ± 6·9164·7 ± 6·4
       Total171·6 ± 9·2172·2 ± 9·5171·0 ± 8·9
      Total N1123601522
      Weight (kg)< 10-6
       Male85·4 ± 15·683·8 ± 14·787·4 ± 15·8
       Female71·7 ± 17·167·8 ± 13·676·8 ± 19·4
       Total78·4 ± 17·674·9 ± 16·282·4 ± 18·4
      Total N1122601521
      Body mass index (kg/m2)< 10-6
       Male26·9 ± 4·525·9 ± 4·028·0 ± 4·7
       Female26·2 ± 6·224·5 ± 5·028·3 ± 7·0
       Total26·5 ± 5·525·2 ± 4·628·1 ± 5·9
      Total N1122601521
      Abbreviations: BNMS, Brisbane Nevus Morphology Study; SD, Standard deviation.
      1 Unaffected, participant with no prior history of melanoma
      2 Case, participant with prior history of at least one melanoma
      3 Unadjusted for other covariates
      4 Face, Shoulder, Hands (none = 0, mild = 1, moderate = 2, severe =3; 0 to 9 total)
      Supplementary Table S2Melanoma OR for Nevus Subtype Count Quintiles
      Bar
      Corresponding bar in Figure 1
      LevelNevus Subtype
      The first quintile for nonspecific nevus count (adjusted to that of a male aged 40 years) was 2.35, for reticular 0.20, and globular 0.34. These correspond to observed counts of 3, 1, and 0 (recalling that the regression adjustment for age can lead to a fractional count).
      Odds Ratio
      Odds ratios for melanoma are relative to the lowest quintile.
      95% CI
      1Q2Globular1.010.60–1.71
      2Q2Reticular1.280.78–2.11
      3Q2Nonspecific1.781.14–2.81
      4Q3Globular1.520.91–2.58
      5Q3Reticular1.781.10–2.89
      6Q3Nonspecific2.371.52–3.74
      7Q4Globular1.460.87–2.49
      8Q4Reticular2.741.69–4.46
      9Q4Nonspecific4.272.73–6.76
      10Q5Globular2.821.72–4.88
      11Q5Reticular2.941.84–4.77
      12Q5Nonspecific7.644.86–12.19
      Abbreviation: CI, confidence interval.
      1 Corresponding bar in Figure 1
      2 The first quintile for nonspecific nevus count (adjusted to that of a male aged 40 years) was 2.35, for reticular 0.20, and globular 0.34. These correspond to observed counts of 3, 1, and 0 (recalling that the regression adjustment for age can lead to a fractional count).
      3 Odds ratios for melanoma are relative to the lowest quintile.
      Supplementary Table S3Logistic Regression Analysis of CM Risk within a Subset of Individuals
      359 cases and 323 controls
      with Separate Counts of Complex and Homogenous Pattern Nevi
      TermOR95%LL95%ULP-value
      (Intercept)0.270.160.43
      Poly (Age, 2) linear2.22x10141.50x1094.762x1019< 2.2x10-16
      Poly (Age, 2) quadratic2.77x1083.50x1032.49x10134.0x10-15
      Sex (M)0.840.581.210.76
      These counts have been divided into quintiles
      Complex Q2 (1,3)
      1.520.892.629.1x10-9
       Q3 (3,6)1.901.113.26
       Q4 (6,14)2.461.424.30
       Q5 (14,112)4.752.409.62
      These counts have been divided into quintiles
      Homogenous Q2 (1,4)
      2.041.183.542.2 x10-16
       Q3 (4,8)2.811.624.93
       Q4 (8,17)3.742.086.81
       Q5 (17,101)11.345.6323.75
      log reticular and globular counts have been treated as continuous predictors
      logReticular
      1.090.871.360.17
      log reticular and globular counts have been treated as continuous predictors
      logGlobular
      0.730.531.000.36
      Abbreviations : CM, cutaneous melanoma; LL, lower limit; M, male; OR, odds ratio; UL, upper limit.
      1 359 cases and 323 controls
      2 These counts have been divided into quintiles
      3 log reticular and globular counts have been treated as continuous predictors
      Supplementary Table S4Logistic Regression Analysis of CM Risk within the BNMS Collection versus Age and Nevus Subtype Counts
      TermBetaSEStatisticOR95%LL95%ULP-value
      (Intercept)-1.700.2-8.390.180.120.270.00
      Age linear12.662.584.913136062012488744780.00
      Age quadratic8.462.593.274728.2529.51757516.220.00
      Counts are log2 transformed, so the ORs represent the increase in risk for each doubling of that nevus count.
      ,
      Missing homogenous and complex counts have been imputed using the R mice package (5 replicates) using age, melanoma status, reticular, globular and total nevus counts.
      Homogeneous
      0.470.076.571.591.391.830.00
      Counts are log2 transformed, so the ORs represent the increase in risk for each doubling of that nevus count.
      ,
      Missing homogenous and complex counts have been imputed using the R mice package (5 replicates) using age, melanoma status, reticular, globular and total nevus counts.
      Complex
      0.310.074.301.361.181.560.00
      Counts are log2 transformed, so the ORs represent the increase in risk for each doubling of that nevus count.
      Globular
      -0.190.11-1.750.830.671.020.08
      Counts are log2 transformed, so the ORs represent the increase in risk for each doubling of that nevus count.
      Reticular
      0.040.080.561.040.901.361.21
      Abbreviations: BNMS, Brisbane Nevus Morphology Study; CM, cutaneous melanoma; LL, lower limit; OR, odds ratio; SE, standard error; UL, upper limit.
      1 Counts are log2 transformed, so the ORs represent the increase in risk for each doubling of that nevus count.
      2 Missing homogenous and complex counts have been imputed using the R mice package (5 replicates) using age, melanoma status, reticular, globular and total nevus counts.
      Supplementary Table S5
      The SNP identified in a bivariate analysis of nevi and melanoma by Duffy et al., (2018), presented in the same order with inclusion of MITF E318K. Given that there are 33 candidate SNPs, the Bonferroni corrected critical threshold for one trait (experiment-wise α < 0.05) would be 0.0015.
      SNP Association with Melanoma, TNC and Nevus Subtypes in 1,235 BNMS Samples
      SNP
      The SNP was directly genotyped unless superscript 3.
      Chr:Position B37Gene/IntervalMelanoma
      Compared with 27,251 individuals directly genotyped in the QIMR Brisbane Longitudinal Twin Study and other studies combined with BNMS controls.
      TNCNonspecificGlobularReticular
      rs869329c9: 21804693MTAP5x10-92x10-67x10-51x10-42.7x10-3
      rs13298522: 38563471PLA2G61x10-41.7x10-30.01396.5x10-30.0296
      rs122035926: 396321IRF44x10-80.02270.44997x10-80.0909
      rs10816595c9: 1107097359q31.23.7x10-30.08460.08940.52210.4129
      rs6009519: 224742DOCK80.03350.37420.11570.13190.2128
      rs731335212: 88949124KITLG0.38310.73100.87180.13600.1747
      rs46708132: 38317710CYP1B10.08210.40030.76880.94800.2213
      rs2514645: 149196234PPARGC1B0.32490.09650.18550.33830.9078
      rs55875066
      The SNP was imputed.
      2: 240076002HDAC40.13140.12560.28820.90240.0215
      rs164087512: 13069524GPRC5A0.01430.69330.67030.42850.2666
      rs4557533810: 5784151FAM208B0.36310.65710.45020.69480.8358
      rs126963043: 169481271TERC0.72760.48960.50870.29920.9124
      rs11764890715: 33277710FMN14x10-110.14120.16542.1x10-30.1302
      rs3446695619: 3353622NFIC0.81020.86490.55580.46550.0150
      rs2357176
      The SNP was imputed.
      14: 64409313SYNE20.72360.02651.0x10-30.25160.3668
      rs1484375
      The SNP was imputed.
      9: 1090675619q31.13.1x10-30.08340.06930.92170.1288
      rs26952371: 226603635PARP10.79060.34630.76990.28070.1421
      rs7300822911: 108187689ATM0.08190.23360.32450.63750.7646
      rs727046581: 150833010SETDB10.70740.24500.16460.65920.9683
      rs1259663816: 54115829FTO0.06290.44830.48870.63380.6469
      rs16367447: 16984280AGR31.2x10-30.25340.19180.60760.9715
      rs41698121: 42745414MX23.7x10-30.72000.79650.29180.1144
      rs7557060416: 89846677MC1R6x10-70.28630.04000.54170.1148
      rs3802865: 1320247TERT3.6x10-37.6x10-34.8x10-30.90690.3737
      rs49813611: 69367118TPCN2/CCND10.63000.04690.06830.93770.0323
      rs75823622: 202176294CASP80.02770.28630.24860.74720.4071
      rs5623868420: 33236696ASIP2x10-90.15010.02270.70660.1221
      rs21255706: 21166705CDKAL10.52630.11710.11230.76390.3278
      rs1083025311: 89028043TYR0.23070.23320.21350.72570.3319
      rs18462847414: 91185865TTC7B0.44010.31370.12330.73050.4347
      rs250417
      The SNP was imputed.
      5: 33952378SLC45A26x10-72x10-54x10-60.03570.1483
      rs477813815: 28335820OCA20.01250.01320.02050.36330.4730
      rs1496179563: 70014091MITF2x10-18
      Compared with 452,264 United Kingdom Biobank controls.
      5x10-42.0 x10-30.045.0x10-3
      Abbreviations: BNMS, Brisbane Nevus Morphology Study; SNP, single nucleotide polymorphism; TNC, total body nevus count.
      1 The SNP identified in a bivariate analysis of nevi and melanoma by
      • Duffy D.L.
      • Zhu G.
      • Li X.
      • Sanna M.
      • Iles M.M.
      • Jacobs L.C.
      • et al.
      Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
      , presented in the same order with inclusion of MITF E318K. Given that there are 33 candidate SNPs, the Bonferroni corrected critical threshold for one trait (experiment-wise α < 0.05) would be 0.0015.
      2 The SNP was directly genotyped unless superscript 3.
      3 The SNP was imputed.
      4 Compared with 27,251 individuals directly genotyped in the QIMR Brisbane Longitudinal Twin Study and other studies combined with BNMS controls.
      5 Compared with 452,264 United Kingdom Biobank controls.
      Supplementary Table S6Source of BNMS Study Participants
      SourceCases

      N (%)
      Unaffected controls

      N (%)
      Total

      N (%)
      Brisbane Longitudinal Twin Study3 (0.49)244 (37.03)247 (19.51)
      QSkin Study2 (0.33)73 (11.08)75 (5.92)
      Melanoma Patients Australia22 (3.62)13 (1.97)35 (2.76)
      Queensland Cancer Registry5 (0.82)05 (0.39)
      Princess Alexandra Hospital Melanoma Surgical Unit, Dermatology Outpatients Department, Oncology Department303 (49.92)79 (11.99)382 (30.17)
      Private dermatology or primary care clinics199 (32.78)29 (4.4)228 (18.01)
      Volunteers - media coverage, advertising or word of mouth58 (9.56)154 (23.37)212 (16.75)
      Volunteers - family members of earlier participants8 (1.32)36 (5.46)44 (3.48)
      Other sources7 (1.15)31 (4.70)48 (3.79)
      Total (100%)6076591266
      Supplementary Table S7Pearson Correlations for Log-Transformed Counts of TNC ≥ 5mm Diameter of Different Dermoscopic Appearance over Body, all Nevi 2-5 mm Diameter on the Back, and of Pigmented Lesions (Freckles and Nevi) of the Iris in Melanoma Cases and Controls: Nevi 2-5 mm Diameter were only Counted on 572 Individuals (213 Cases, 334 Controls) (
      • Laino A.M.
      • Berry E.G.
      • Jagirdar K.
      • Lee K.J.
      • Duffy D.L.
      • Soyer H.P.
      • et al.
      Iris pigmented lesions as a marker of cutaneous melanoma risk: an Australian case-control study.
      )
      Log numberTNC ≥ 5mmNon-specific ≥ 5mmGlobular ≥ 5mmReticular ≥ 5mmBack nevi 2-5 mmIris pigmented lesions
      TNC ≥ 5mm1
      Non-specific ≥ 5mm0.941
      Globular ≥ 5mm0.500.331
      Reticular ≥ 5mm0.500.300.451
      Back nevi 2-5 mm0.610.500.500.701
      Iris pigmented lesions0.220.210.020.090.171
      Abbreviation: TNC, total body nevus count.
      Supplementary Table S8Hair and Eye Color in BNMS Controls, Compared to Large Australian Population-Based Studies
      FemalesMales
      self-reported hair, observer eye color.
      BNMS Control N (%)
      QSkin: QSkin Sun and Health Study, self-reported hair and eye color.
      QSkin N (%)
      observer hair and eye color.
      BMES N (%)
      self-reported hair, observer eye color.
      BNMS Control N (%)
      QSkin: QSkin Sun and Health Study, self-reported hair and eye color.
      QSkin N (%)
      observer hair and eye color.
      BMES N (%)
      Hair Color
       Black14 (3.9)1394 (5.9)155 (7.7)17 (5.8)2971 (15.0)169 (10.9)
       Dark Brown105 (28.9)8195 (34.6)1251 (61.8)112 (38.5)6313 (31.8)973 (62.8)
       Light Brown145 (39.9)8910 (37.6)95 (32.6)7186 (36.2)
       Blonde/Fair75 (20.7)3723 (15.7)454 (22.4)44 (15.1)2366 (11.9)319 (20.6)
       Red24 (6.6)1482 (6.3)165 (8.1)23 (7.9)993 (5.0)88 (5.7)
      Total (100%)36323,704202529119,8291549
      Eye Color
       Blue/Gray184 (50.7)8227 (35.2)856 (45.6)144 (56.8)8110 (41.4)821 (56.1)
       Green/Hazel102 (28.1)9479 (40.6)579 (30.8)85 (26.5)6613 (33.8)362 (24.8)
       Brown/Black77 (21.2)5669 (24,3)445 (23.6)64 (16.7)4835 (24.7)280 (19.1)
      Total36323,375188029319,5581463
      Abbreviations: BMES, Blue Mountains Eye Study; BNMS, Brisbane Naevus Morphology Study.
      1 self-reported hair, observer eye color.
      2 QSkin: QSkin Sun and Health Study, self-reported hair and eye color.
      3 observer hair and eye color.
      Supplementary Table S9Association of Chromosomal Regions and SNPs with TNC and Nevus Subtype Counts in BNMS
      Association P-value, highest for TNC and includes the highest nevus subtype SNP where relevant.
      Phenotypechrrs numberpositionAllele 1Allele 0Freq

      Allele 1
      BetaStandard errorP-valueLoci within region
      Reticular1rs497061239125390CT0.218-0.1190.02272.06e-07RP11=334L9.1

      RRAGC

      MYCBP

      RHBDL2
      TNC1rs497061239125390CT0.218-0.1190.02272.06e-07RP11=334L9.1

      RRAGC

      MYCBP

      RHBDL2

      RPS7P5
      TNC1rs7545703240167654CT0.6290.1030.01961.67e-07
      Non-specific1rs7545703240167654CT0.6290.0970.01958.03e-07RPS7P5
      Reticular1rs7545703240167654CT0.6290.1040.01961.67e-07
      Globular2indel.591629547629DI0.0260.2890.04148.51e-12
      Abbreviations: BNMS, Brisbane Nevus Morphology Study; SNP, single nucleotide polymorphism; TNC, total body nevus count.
      1 Association P-value, highest for TNC and includes the highest nevus subtype SNP where relevant.
      Supplementary Table S10MTAP, PLA2G6, and IRF4 SNP Associations with TNC and Nevus Subtypes in 1,235 BNMS Samples Calculated Using the GEMMA Package
      Gene/SNPTotal BodyGlobularReticularNonspecific
      BetaSEp-valueBetaSEp-valueBetaSEp-valueBetaSEp-value
      MTAP rs7023329*A8.07x10-21.90x10-22.3x10-55.24x10-21.48x10-24.1x10-48.06x10-21.90 x10-22.31x10-56.95x10-21.89x10-22.3x10-4
      PLA2G6 rs132985*C5.89x10-21.88x10-21.7x10-34.32x10-21.46x10-23.13x10-35.89x10-21.88x10-21.7x10-34.64x10-21.87x10-21.3x10-2
      IRF4 rs12203592*C5.69x10-22.26x10-21.1x10-29.55x10-21.74x10-26.04x10-85.69x10-22.25x10-21.1x10-21.82x10-22.25x10-24.1x10-1
      Abbreviations: BNMS, Brisbane Nevus Morphology Study; SNP, single nucleotide polymorphism; TNC, total body nevus count.
      Supplementary Table S11MC1R Variant Genotype Association with Melanoma, TNC and Nevus Subtypes in 1,235 BNMS Samples
      MC1R GenotypeCM
      Odds ratios for CM are from a logistic regression model adjusting for age, age squared, sex and four ancestry principal components.
      OR (95% CI)
      TNC
      Nevus counts are predicted counts for that MC1R genotype for a 60 year aged male control of study average ancestry.
      (95% CI)
      Nonspecific (95% CI)Globular (95% CI)Reticular (95% CI)
      WT/WT1.008.9 [7.3–10.7]5.6 [4.6–6.9]0.6 [0.4–0.8]2.0 [1.5–2.5]
      WT/r2.1 [1.4–3.2]10.0 [8.3–12]6.9 [5.6–8.3]0.7 [0.5– 0.9]1.9 [1.5–2.4]
      r/r3.0 [1.7–5.2]10.2 [7.9–13.2]7.5 [5.7–9.7]0.4 [0.2–0.7]2.0 [1.4 –2.7]
      WT/R2.9 [1.9–4.4]10.5 [8.7–12.7]7.7 [6.3–9.4]0.6 [0.4–0.8]1.7 [1.3– 2.2]
      R/r4.0 [2.6–6.2]14.2 [11.7–17.3]11.3 [9.2–13.8]0.6 [0.4–0.8]1.8 [1.4 –2.3]
      R/R5.6 [3.2–9.9]9.9 [7.5–12.8]8.9 [6.8–11.6]0.4 [0.1–0.6]1.0 [0.6–1.5]
      P-value
      Association P-value from linear model of log-transformed nevus count adjusting for age, age squared, sex and CM status, and four ancestry principal components.
      9.0x10-123.9x10-41.4x10-80.170.015
      Abbreviations: BNMS, Brisbane Nevus Morphology Study; CI, confidence interval; CM, cutaneous melanoma; OR, odds ratio; SNP, single nucleotide polymorphism; TNC, total body nevus count.
      1 Odds ratios for CM are from a logistic regression model adjusting for age, age squared, sex and four ancestry principal components.
      2 Nevus counts are predicted counts for that MC1R genotype for a 60 year aged male control of study average ancestry.
      3 Association P-value from linear model of log-transformed nevus count adjusting for age, age squared, sex and CM status, and four ancestry principal components.
      Supplementary Table S12IRF4 Variant Genotype Interaction with Melanoma Status, TNC, and Nevus Subtypes
      Tabulated counts are those predicted for a 60-year-old man with study-average values for the four ancestry principal components from a linear model for log-transformed nevus counts, including age, age squared, sex, and Principal components fitted separately to melanoma cases and controls.
      CM statusrs12203592TNC
      Difference between genotypic means for cases, P-value = 0.003
      ,
      Interaction P-value = 0.005. The full regression model for total body nevus count ≥ 5mm diameter including an interaction term for melanoma case and IRF4 genotype finds that the difference between cases and controls is statistically significant, so the rs12203592*T allele decreases total body nevus count in cases only.
      (95% CI)
      Nonspecific (95% CI)Globular (95% CI)Reticular (95% CI)
      CaseC/C25.3 [21.7–29.6]16.1 [13.5–19]1.6 [1.3–1.9]3.7 [3–4.5]
      C/T19.5 [16.2–23.4]13.7 [11.2–6.7]0.8 [0.6–1.1]3.0 [2.3–3.8]
      T/T15.8 [11.4 – 21.8]18.7 [13.3–26.1]0.5 [0.2–1]2.2 [1.3–3.4]
      ControlC/C10.3 [8.4–12.5]7.2 [5.9–8.8]0.7 [0.5–1]1.6 [1.3–2]
      C/T11.3 [9–14.1]8.4 [6.7–10.5]0.6 [0.4–0.9]1.8 [1.4–2.3]
      T/T9.6  [6.2–14.6]8.2 [5.3–12.4]0.3 [0–0.7]1.4 [0.8–2.3]
      Abbreviations: CI, confidence interval; CM, cutaneous melanoma; TNC, total body nevus count.
      1 Tabulated counts are those predicted for a 60-year-old man with study-average values for the four ancestry principal components from a linear model for log-transformed nevus counts, including age, age squared, sex, and Principal components fitted separately to melanoma cases and controls.
      2 Difference between genotypic means for cases, P-value = 0.003
      3 Interaction P-value = 0.005. The full regression model for total body nevus count ≥ 5mm diameter including an interaction term for melanoma case and IRF4 genotype finds that the difference between cases and controls is statistically significant, so the rs12203592*T allele decreases total body nevus count in cases only.
      Supplementary Table S13Model Results for IRF4 Variant Genotype Interaction with Melanoma Status, TNC, and Nevus Subtypes
      Model TermEstimateStd. ErrorF valueP value
      (Intercept)0.27961470.1638953
      Poly (Age, 2) linear1.52569290.853548682.5622< 2.2e-16***
      Poly (Age, 2) quadratic-4.36192150.7495693
      Male Sex0.06850300.026140611.23360.0008278***
      Ancestry PC1-0.00111700.001378416.68904.690e-05***
      Ancestry PC20.00325720.000919817.99812.378e-05***
      Ancestry PC30.00265450.00234222.16250.1416695
      Ancestry PC4-0.00294050.00174993.53280.0604037
      Melanoma case0.38985240.0350442131.5341< 2.2e-16***
      rs12203592*T0.01354930.03007655.14150.0235343*
      Case:rs12203592*T-0.11807550.04178797.98390.0047961**
      Abbreviations: PC, principal component; Std, standard; TNC, total body nevus count. *P < 0.05; **P < 0.01; ***P < 0.001.
      The full regression model for total body nevus count ≥ 5mm diameter. including an interaction term for melanoma case and IRF4 genotype finds that the difference between cases and controls is statistically significant, so the rs12203592*T allele decreases the total body nevus count in cases only.

      References

        • Bajaj S.
        • Dusza S.W.
        • Marchetti M.A.
        • Wu X.
        • Fonseca M.
        • Kose K.
        • et al.
        Growth-curve modeling of nevi with a peripheral globular pattern.
        JAMA Dermatol. 2015; 151: 1338-1345
        • Bataille V.
        • Bishop J.A.
        • Sasieni P.
        • Swerdlow A.J.
        • Pinney E.
        • Griffiths K.
        • et al.
        Risk of cutaneous melanoma in relation to the numbers, types and sites of naevi: a case-control study.
        Br J Cancer. 1996; 73: 1605-1611
        • Blake T.
        • McClenahan P.
        • Duffy D.
        • Schaider H.
        • McEniery E.
        • Soyer H.P.
        Distribution analyses of acquired melanocytic naevi on the trunk.
        Dermatol Basel Switzerland. 2014; 228: 269-275
        • Duffy D.L.
        • Iles M.M.
        • Glass D.
        • Zhu G.
        • Barrett J.H.
        • Höiom V.
        • et al.
        IRF4 variants have age-specific effects on nevus count and predispose to melanoma.
        Am J Hum Genet. 2010; 87: 6-16
        • Duffy D.L.
        • Zhu G.
        • Li X.
        • Sanna M.
        • Iles M.M.
        • Jacobs L.C.
        • et al.
        Novel pleiotropic risk loci for melanoma and nevus density implicate multiple biological pathways.
        Nat Commun. 2018; 9: 4774
        • Fonseca M.
        • Marchetti M.A.
        • Chung E.
        • Dusza S.W.
        • Burnett M.E.
        • Marghoob A.A.
        • et al.
        Cross-sectional analysis of the dermoscopic patterns and structures of melanocytic naevi on the back and legs of adolescents.
        Br J Dermatol. 2015; 173: 1486-1493
        • Law M.H.
        • Bishop D.T.
        • Lee J.E.
        • Brossard M.
        • Martin N.G.
        • Moses E.K.
        • et al.
        Genome-wide meta-analysis identifies five new susceptibility loci for cutaneous malignant melanoma.
        Nat Genet. 2015; 47: 987-995
        • Lee S.
        • Duffy D.L.
        • McClenahan P.
        • Lee K.J.
        • McEniery E.
        • Burke B.
        • et al.
        Heritability of naevus patterns in an adult twin cohort from the Brisbane Twin Registry: a cross-sectional study.
        Br J Dermatol. 2016; 174: 356-363
        • Marchetti M.A.
        • Kiuru M.H.
        • Busam K.J.
        • Marghoob A.A.
        • Scope A.
        • Dusza S.W.
        • et al.
        Melanocytic naevi with globular and reticular dermoscopic patterns display distinct BRAF V600E expression profiles and histopathological patterns.
        Br J Dermatol. 2014; 171: 1060-1065
        • McWhirter S.R.
        • Duffy D.L.
        • Lee K.J.
        • Wimberley G.
        • McClenahan P.
        • Ling N.
        • et al.
        Classifying dermoscopic patterns of naevi in a case-control study of melanoma.
        PLOS ONE. 2017; 12: e0186647
        • Orlow I.
        • Satagopan J.M.
        • Berwick M.
        • Enriquez H.L.
        • White K.A.
        • Cheung K.
        • et al.
        Genetic factors associated with naevus count and dermoscopic patterns: preliminary results from the Study of Nevi in Children (SONIC).
        Br J Dermatol. 2015; 172: 1081-1089
        • Ribero S.
        • Zugna D.
        • Osella-Abate S.
        • Glass D.
        • Nathan P.
        • Spector T.
        • et al.
        Prediction of high naevus count in a healthy UK population to estimate melanoma risk.
        Br J Dermatol. 2016; 174: 312-318
        • Suh K.Y.
        • Bolognia J.L.
        Signature nevi.
        J Am Acad Dermatol. 2009; 60: 508-514
        • Tan J.M.
        • Tom L.N.
        • Soyer H.P.
        • Stark M.S.
        Defining the Molecular Genetics of dermoscopic naevus patterns.
        Dermatol Basel Switzerland. 2019; 235: 19-34
        • Zalaudek I.
        • Schmid K.
        • Marghoob A.A.
        • Scope A.
        • Manzo M.
        • Moscarella E.
        • et al.
        Frequency of dermoscopic nevus subtypes by age and body site: a cross-sectional study.
        Arch Dermatol. 2011; 147: 663-670