Portal de Periódicos da CAPES

BRAF Polymorphisms and Risk of Melanocytic Neoplasia

2005; Elsevier BV; Volume: 125; Issue: 6 Linguagem: Inglês

10.1111/j.0022-202x.2005.23937.x

1523-1747

Michael R. James, Richard B. Roth, Michael Shi, Stefan Kammerer, Matthew R. Nelson, Mitchell Stark, Troy Dumenil, Grant W. Montgomery, Nicholas K. Hayward, Nicholas G. Martin, Andreas Braun, David L. Duffy,

Computational Drug Discovery Methods

Somatic mutations of the BRAF gene are common in melanomas and nevi but the contribution of polymorphisms in this gene to melanoma or nevus susceptibility remains unclear. An Australian melanoma case–control sample was typed for 16 single nucleotide polymorphisms (SNP) within the BRAF gene, and five SNP in three neighboring genes. The sample comprised 755 melanoma cases from 740 families stratified by family history of melanoma and controls from 635 unselected twin families (2239 individuals). Ancestry of the cases and controls was recorded, and the twins had undergone skin examination to assess total body nevus count, degree of freckling, and pigmentation phenotype. Genotyping was carried out via primer extension followed by matrix-assisted laser desorption ionization-time of flight mass spectrometry. SNP in the BRAF gene were found to be weakly associated with melanoma status but not with development of nevi or freckles. The estimated proportion of attributable risk of melanoma due to variants in BRAF is 1.6%. This study shows that BRAF polymorphisms predispose to melanoma but the causal variant has yet to be determined. The burden of disease associated with this variant is greater than that associated with the major melanoma susceptibility locus CDKN2A, which has an estimated attributable risk of 0.2%. Somatic mutations of the BRAF gene are common in melanomas and nevi but the contribution of polymorphisms in this gene to melanoma or nevus susceptibility remains unclear. An Australian melanoma case–control sample was typed for 16 single nucleotide polymorphisms (SNP) within the BRAF gene, and five SNP in three neighboring genes. The sample comprised 755 melanoma cases from 740 families stratified by family history of melanoma and controls from 635 unselected twin families (2239 individuals). Ancestry of the cases and controls was recorded, and the twins had undergone skin examination to assess total body nevus count, degree of freckling, and pigmentation phenotype. Genotyping was carried out via primer extension followed by matrix-assisted laser desorption ionization-time of flight mass spectrometry. SNP in the BRAF gene were found to be weakly associated with melanoma status but not with development of nevi or freckles. The estimated proportion of attributable risk of melanoma due to variants in BRAF is 1.6%. This study shows that BRAF polymorphisms predispose to melanoma but the causal variant has yet to be determined. The burden of disease associated with this variant is greater than that associated with the major melanoma susceptibility locus CDKN2A, which has an estimated attributable risk of 0.2%. Brisbane Twin Nevus Study cutaneous malignant melanoma Queensland Familial Melanoma Project single nucleotide polymorphism The field of cancer genetics has produced notable successes applying genetic linkage analysis to identify genes that confer significant inherited susceptibility to particular cancers that show specific familial clustering. So far, these have been variants in genes of relatively large effect but which are rare in the general population, e.g., BRCA1 and BRCA2 in breast cancer and CDKN2A in melanoma, and thus do not account for a major contribution to overall population risks. Unlike these prominent cases it seems likely that most common cancers with an inherited component involve more genes, each conferring a fraction of the risk, and perhaps particular variants in multiple genes acting epistatically. Clearly, these are not readily amenable to discovery using the linkage approach and instead an association design may be more successful (Barrett et al., 2005Barrett J.C. Fry B. Maller J. Daly M.J. Haploview analysis and visualization of LD and haplotype maps.Bioinformatics. 2005; 23: 263-265Crossref Scopus (12023) Google Scholar). Thus far, such association studies have had mixed results, arguably due to inadequate sample size and suboptimal study design (Risch, 2001Risch N. The genetic epidemiology of cancer: Interpreting family and twin studies and their implications for molecular genetic approaches.Cancer Epidemiol Biomark Prev. 2001; 4: 469-473Google Scholar). Here, we present a large association study of the role in genetic predisposition to melanoma susceptibility of the BRAF gene, in a systematic and comprehensive haplotyping approach across the entire locus and adjacent genes. The Ras/Raf/MAPK (mitogen-activated protein kinase) pathway is a critical molecular signaling cascade through which extracellular signals can be transmitted into the nucleus to regulate cell proliferation or differentiation through altered gene expression. Constitutive activation of this pathway is a frequent event in melanoma development (Cohen et al., 2002Cohen C. Zavala-Pompa A. Sequeira J.H. et al.Mitogen-activated protein kinase activation is an early event in melanoma progression.Clin Cancer Res. 2002; 8: 3728-3733PubMed Google Scholar; Dong et al., 2003Dong J. Phelps R.G. Qiao R. Yao S. Benard O. Ronai Z. Aaronson S.A. BRAF oncogenic mutations correlate with progression rather than initiation of human melanoma.Cancer Res. 2003; 63: 3883-3885PubMed Google Scholar; Satyamoorthy et al., 2003Satyamoorthy K. Li G. Gerrero M.R. et al.Constitutive mitogen-activated protein kinase activation in melanoma is mediated by both BRAF mutations and autocrine growth factor stimulation.Cancer Res. 2003; 63: 756-759PubMed Google Scholar). Recently, BRAF gene somatic mutations have been shown to be associated with malignant melanoma (Davies et al., 2002Davies H. Bignell G.R. Cox C. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar), being present in 40%–88% of cutaneous melanomas (Brose et al., 2002Brose M.S. Volpe P. Feldman M. et al.BRAF and RAS mutations in human lung cancer and melanoma.Cancer Res. 2002; 62: 6997-7000PubMed Google Scholar; Davies et al., 2002Davies H. Bignell G.R. Cox C. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar; Dong et al., 2003Dong J. Phelps R.G. Qiao R. Yao S. Benard O. Ronai Z. Aaronson S.A. BRAF oncogenic mutations correlate with progression rather than initiation of human melanoma.Cancer Res. 2003; 63: 3883-3885PubMed Google Scholar; Gorden et al., 2003Gorden A. Osman I. Gai W. et al.Analysis of BRAF and N-RAS mutations in metastatic melanoma tissues.Cancer Res. 2003; 63: 3955-3957PubMed Google Scholar; Kumar et al., 2003Kumar R. Angelini S. Czene K. Sauroja I. Hahka-Kemppinen M. Pyrhonen S. Hemminki K. BRAF mutations in metastatic melanoma: A possible association with clinical outcome.Clin Cancer Res. 2003; 9: 3362-3368PubMed Google Scholar; Pollock et al., 2003Pollock P.M. Harper U.L. Hansen K.S. et al.High frequency of BRAF mutations in nevi.Nat Genet. 2003; 33: 19-20Crossref PubMed Scopus (1359) Google Scholar; Satyamoorthy et al., 2003Satyamoorthy K. Li G. Gerrero M.R. et al.Constitutive mitogen-activated protein kinase activation in melanoma is mediated by both BRAF mutations and autocrine growth factor stimulation.Cancer Res. 2003; 63: 756-759PubMed Google Scholar), whereas being essentially absent in control tissues. Mutations are also extremely common (74%–82%) in benign melanocytic nevi (Pollock et al., 2003Pollock P.M. Harper U.L. Hansen K.S. et al.High frequency of BRAF mutations in nevi.Nat Genet. 2003; 33: 19-20Crossref PubMed Scopus (1359) Google Scholar; Yazdi et al., 2003Yazdi A.S. Palmedo G. Flaig M.J. Kutzner H. Sander C.A. SP-11 Different frequencies of a BRAF point mutation in melanocytic skin lesions.Pigment Cell Res. 2003; 16: 580Crossref Google Scholar), an observation arguing for a critical role for B-Raf in initiating melanocytic neoplasia. All somatic mutations documented to date in melanoma have been found in the kinase domain of B-Raf, encoded by exons 11 and 15 of the BRAF gene (Brose et al., 2002Brose M.S. Volpe P. Feldman M. et al.BRAF and RAS mutations in human lung cancer and melanoma.Cancer Res. 2002; 62: 6997-7000PubMed Google Scholar; Davies et al., 2002Davies H. Bignell G.R. Cox C. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar; Pollock et al., 2003Pollock P.M. Harper U.L. Hansen K.S. et al.High frequency of BRAF mutations in nevi.Nat Genet. 2003; 33: 19-20Crossref PubMed Scopus (1359) Google Scholar; Satyamoorthy et al., 2003Satyamoorthy K. Li G. Gerrero M.R. et al.Constitutive mitogen-activated protein kinase activation in melanoma is mediated by both BRAF mutations and autocrine growth factor stimulation.Cancer Res. 2003; 63: 756-759PubMed Google Scholar; Casula et al., 2004Casula M. Colombino M. Satta M.P. et al.BRAF gene is somatically mutated but does not make a major contribution to malignant melanoma susceptibility: The Italian melanoma intergroup study.J Clin Oncol. 2004; 22: 286-292Crossref PubMed Scopus (47) Google Scholar) (Figure 1). The vast majority (over 90%) of these mutations affect codon 599 and result in a valine to glutamic acid substitution, see note added in proof which is thought to lead to constitutive activation of Ras/Raf/MAPK signal transduction (Davies et al., 2002Davies H. Bignell G.R. Cox C. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar). Other less frequent BRAF coding somatic mutations in melanoma are G468A in exon 11 and L596V and Q608H in exon 15 (Davies et al., 2002Davies H. Bignell G.R. Cox C. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar; Pollock et al., 2003Pollock P.M. Harper U.L. Hansen K.S. et al.High frequency of BRAF mutations in nevi.Nat Genet. 2003; 33: 19-20Crossref PubMed Scopus (1359) Google Scholar; Casula et al., 2004Casula M. Colombino M. Satta M.P. et al.BRAF gene is somatically mutated but does not make a major contribution to malignant melanoma susceptibility: The Italian melanoma intergroup study.J Clin Oncol. 2004; 22: 286-292Crossref PubMed Scopus (47) Google Scholar). In some melanomas without BRAF mutations, the MAPK pathway is constitutively activated through mutation of NRAS (van Elsas et al., 1996van Elsas A. Zerp S.F. van der Flier S. et al.Relevance of ultraviolet-induced N-ras oncogene point mutations in development of primary human cutaneous melanoma.Am J Pathol. 1996; 149: 883-893PubMed Google Scholar). BRAF and NRAS mutations appear to have similar effects in melanoma development since their presence in any single tumor is mutually exclusive (Brose et al., 2002Brose M.S. Volpe P. Feldman M. et al.BRAF and RAS mutations in human lung cancer and melanoma.Cancer Res. 2002; 62: 6997-7000PubMed Google Scholar; Davies et al., 2002Davies H. Bignell G.R. Cox C. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar; Pollock et al., 2003Pollock P.M. Harper U.L. Hansen K.S. et al.High frequency of BRAF mutations in nevi.Nat Genet. 2003; 33: 19-20Crossref PubMed Scopus (1359) Google Scholar; Satyamoorthy et al., 2003Satyamoorthy K. Li G. Gerrero M.R. et al.Constitutive mitogen-activated protein kinase activation in melanoma is mediated by both BRAF mutations and autocrine growth factor stimulation.Cancer Res. 2003; 63: 756-759PubMed Google Scholar). This situation is similar in lung (Brose et al., 2002Brose M.S. Volpe P. Feldman M. et al.BRAF and RAS mutations in human lung cancer and melanoma.Cancer Res. 2002; 62: 6997-7000PubMed Google Scholar; Davies et al., 2002Davies H. Bignell G.R. Cox C. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar; Naoki et al., 2002Naoki K. Chen T.H. Richards W.G. Sugarbaker D.J. Meyerson M. Missense mutations of the BRAF gene in human lung adenocarcinoma.Cancer Res. 2002; 62: 7001-7003PubMed Google Scholar), colon (Davies et al., 2002Davies H. Bignell G.R. Cox C. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar; Rajagopalan et al., 2002Rajagopalan H. Bardelli A. Lengauer C. Kinzler K.W. Vogelstein B. Velculescu V.E. Tumorigenesis: RAF/RAS oncogenes and mismatch-repair status.Nature. 2002; 418: 934Crossref PubMed Scopus (1065) Google Scholar; Yuen et al., 2002Yuen S.T. Davies H. Chan T.L. et al.Similarity of the phenotypic patterns associated with BRAF and KRAS mutations in colorectal neoplasia.Cancer Res. 2002; 62: 6451-6455PubMed Google Scholar), and thyroid cancers (Kimura et al., 2003Kimura E.T. Nikiforova M.N. Zhu Z. Knauf J.A. Nikiforov Y.W. Fagin J.A. High prevalence of BRAF mutations in thyroid cancer: Genetic evidence for constitutive activation of the RET/PTC-RAS-BRAF signaling pathway in papillary thyroid carcinoma.Cancer Res. 2003; 63: 1454-1457PubMed Google Scholar), where BRAF and RAS mutations are rarely found together. To date, only one common germline BRAF coding variant has been reported, a synonymous G642G change (rs1042179 (National Center for Biotechnology Information (NCBI), 2003National Center for Biotechnology Information (NCBI) Single nucleotide polymorphism database.Build. 2003; 118http://www.ncbi.nlm.nih.gov/dbSNPGoogle Scholar)). Although this particular variant was not tested,Meyer et al., 2003bMeyer P. Sergi C. Garbe C. Polymorphisms of the BRAF gene predispose males to malignant melanoma.J Carcinogen. 2003; 2: 7Crossref PubMed Scopus (39) Google Scholar recentlydescribed a case–control study (502 melanoma cases; 450 controls) assessing possible associations between intronic single nucleotide polymorphisms (SNP) in the BRAF gene and melanoma risk. Six of 12 SNP analyzed in this German case–control study were significantly (p G0.2051.290.525BRAFB1rs765373PromoterC>T0.1384.320.116B2rs78107575′UTRA>G0.0731.320.517B3rs1267621Intron-1G>A0.1335.410.067B4rs1267606Intron-2T>G0.0614.240.120B5rs1267601Intron-2A>G0.0619.790.007B6rs1267609Intron-3G>A0.0607.020.030B7rs1267649Intron-5C>G0.0617.020.030B8rs1639675Intron-7T>C0.0619.070.011B9rs1267636Intron-8A>G0.0616.880.032B10rs1267646Intron-9C>T0.1354.170.124B11rs1639679Intron-11C>A0.0607.110.029B12IVS12-48CTIntron-12C>T0.0622.590.274B13rs1267639Intron-13C>T0.0629.040.011B14rs1267638Intron-13T>C0.0626.970.031B15rs1042179Exon-16A>G0.1406.60.037B16rs3789806Intron-16C>G0.1416.710.035ADCK2A1rs1046515C>T0.0616.630.036Q9ULE3Q1rs269238T>G0.2903.390.184Q2rs3748088C>T0.3374.160.125Q3rs4726882A>C0.5001.580.453SNP, single nucleotide polymorphism.a Within the 18 exon transcript ENST00000288602.b Minor allele frequency in controls. Open table in a new tab SNP, single nucleotide polymorphism. The B5 (rs1267621) SNP minor allele frequency was highest in the high familial risk cases (Table I), and a test for trend in family risk was significant (p=0.008). In the 10 cases (from eight high-risk families) carrying a mutation in CDKN2A, the major familial melanoma predisposition gene, none carried the rare allele (Table II).Table IIAssociation analysis. Increased gradient of BRAF association with familial risk of melanomaPopulation groupC/CC/AA/AMinor allele frequencyp-valueControls (n=1170)0.8740.1230.0030.064Low familial risk (n=448)0.8590.1270.0130.0770.041Intermediate familial risk (n=150)0.9000.0730.0270.0630.002High familial risk (n=59)0.7800.2200.0000.1100.099CDKN2A mutation families (n=8)1.0000.0000.0000.0000.614Genotypic frequencies of the B2 SNP in the Australian case–control sample stratified by familial risk class and CDKN2A mutation status. SNP, single nucleotide polymorphism. Open table in a new tab Genotypic frequencies of the B2 SNP in the Australian case–control sample stratified by familial risk class and CDKN2A mutation status. SNP, single nucleotide polymorphism. The 16 BRAF SNP were in extremely tight linkage disequilibrium (LD), such that D′ values for pairs of adjacent SNP ranged from 0.72 to 1 (Figure 2). This degree of LD results in only three haplotypes defining 98% of chromosomes (Table III). The strength of association of the common haplotypes combining the BRAF SNP was not greater than that evidenced by the individual SNP with melanoma (Table IV). For example, SNP B5, or equivalently, the “C” haplotype, accounted for a proportion attributable risk estimated as 1.6% (95% CI=0.1%–3.1%). The histological level and Breslow thickness of tumors were not correlated with genotype, either for individual SNP or overall haplotype (Table IV).Table IIIFrequency of BRAF haplotypes and association with CMM risk, adjusted for mole count, hair, and skin colorHaplotypeaSNP numbering according to Table I (1=B1, etc.).FrequencyOdds ratio (95% CI)p-value12345678910111213141516ACAGTAGCTACACCTAC0.8541.00BTGATAGCTATATCTGG0.0721.07 (0.78–1.46)0.68CTAAGGAGCGTCCTCGG0.0591.24 (0.89–1.73)0.21Other––––––––––––––––0.0161.13 (0.461–2.75)0.79CMM, cutaneous malignant melanoma; SNP, single nucleotide polymorphism.a SNP numbering according to Table I (1=B1, etc.). Open table in a new tab Table IVAssociation of BRAF haplotypes with melanoma, and with tumor thickness (mm) among casesBRAF genotypeaBased on haplotypes given in Table III.Odds ratio (95% CI)Breslow thickness (95% CI)A/A1.00 (reference)0.61 (0.54–0.69)A/B1.07 (0.80–1.42)0.69 (0.52–0.86)A/C0.94 (0.69–1.29)0.61 (0.43–0.78)B/B0.71 (0.15–2.57)0.74 (0.00–1.85)B/C1.24 (0.41–3.60)0.24 (0.00–0.94)C/C5.80 (1.40–39.07)0.62 (0.03–1.21)a Based on haplotypes given in Table III. Open table in a new tab CMM, cutaneous malignant melanoma; SNP, single nucleotide polymorphism. There was no association between any of the BRAF SNP and counts of total, macular, papular, or atypical nevi in the Australian controls (best p-value=0.16; see Table V). Similarly, no association was seen with freckling (Table V). Numbers of genotypes containing the “C” haplotype were not large, however.Table VAssociation of BRAF haplotypes with nevus counts and freckling scores in adolescent twin controlsaNo differences between genotypes are significant at the 5% level.BRAF genotypebBased on haplotypes given in Table III.Number genotypedMedian total nevus count (95% CI)Median papular nevus count (95% CI)Mean freckling score (95%CI)A/A822106 (101.9–110.1)17 (15.8–18.1)2.6 (2.4–2.8)A/B144102 (92.0–111.9)17 (14.6–19.4)2.6 (2.1–3.1)A/C110104 (94.0–113.9)19.5 (16.6–22.4)2.4 (1.9–3.0)B/B7682 (63.2–100.8)7 (1.6–12.4)0.5 (-3.0 to 4.0)B/C1268 (28.1–107.9)10.5 (5.0–16.0)4.1 (2.4–5.9)C/C1242412 (-3.0 to 7.0)a No differences between genotypes are significant at the 5% level.b Based on haplotypes given in Table III. Open table in a new tab The question of whether germline coding region mutations of the BRAF gene are responsible for predisposition in large multiple-case melanoma families has been investigated by several groups. The answer appears to be that such large impact mutations are not responsible for any significant proportion of familial or sporadic melanoma since no germline exon 15 BRAF mutations have been found after screening of 42 familial and two sporadic melanoma cases (Lang et al., 2003Lang J. Boxer M. MacKie R. Absence of exon 15 BRAF germline mutations in familiar melanoma.Hum Mutat. 2003; 21: 327-330Crossref PubMed Scopus (58) Google Scholar); 46 familial melanoma cases, 21 multiple melanoma cases, and 106 sporadic melanoma cases (Meyer et al., 2003aMeyer P. Klaes R. Schmitt C. Boettger M.B. Garbe C. Exclusion of BRAFV599E as a melanoma susceptibility mutation.Int J Cancer. 2003; 106: 78-80Crossref PubMed Scopus (34) Google Scholar); 35 familial melanoma cases, 16 multiple melanoma cases, 18 uveal melanoma cases, and 11 probands from familes with melanoma nervous system tumors (Laud et al., 2003Laud K. Kannengiesser C. Avril M.F. et al.BRAF as a melanoma susceptibility candidate gene?.Cancer Res. 2003; 63: 3061-3065PubMed Google Scholar), respectively. Most recently, however, a germline variant, Q608H, in exon 15 has been found (Casula et al., 2004Casula M. Colombino M. Satta M.P. et al.BRAF gene is somatically mutated but does not make a major contribution to malignant melanoma susceptibility: The Italian melanoma intergroup study.J Clin Oncol. 2004; 22: 286-292Crossref PubMed Scopus (47) Google Scholar; Casula et al., 2005Casula M. Colombino M. Satta M.P. et al.Errata to: BRAF gene is somatically mutated but does not make a major contribution to malignant melanoma susceptibility: The Italian melanoma intergroup study.J Clin Oncol. 2005; 23: 936Crossref Google Scholar) though its functional significance is unknown. Nevertheless, as 569 CMM cases were screened, the low incidence (0.18%) of these mutations cannot play a major role in melanoma susceptibility. In a similar vein, a recent study byLaud et al., 2003Laud K. Kannengiesser C. Avril M.F. et al.BRAF as a melanoma susceptibility candidate gene?.Cancer Res. 2003; 63: 3061-3065PubMed Google Scholar suggested that BRAF is also not a low-risk susceptibility gene for melanoma. This group screened the entire coding region of BRAF for germline mutations in melanoma-prone families and sporadic cases (n=80). They found 13 variants (four silent exonic and nine intronic), but none segregated in melanoma-prone families, and there were no significant differences in heterozygote frequencies between cases and controls for any of the 13 SNP. In stark contrast, the results of our study and that ofMeyer et al., 2003bMeyer P. Sergi C. Garbe C. Polymorphisms of the BRAF gene predispose males to malignant melanoma.J Carcinogen. 2003; 2: 7Crossref PubMed Scopus (39) Google Scholar suggest that BRAF is indeed a low-risk susceptibility gene for melanoma. There are several plausible reasons for the discrepant results found between these studies. Laud et al analyzed a much smaller number of samples (n=80) compared to our study, in which 753 cases were utilized. The larger data set provides more power to detect SNP frequency differences, particularly if allele frequency differences are small yet significant. In addition, for each SNP, Laud et al calculated heterozygote frequencies and determined that there were no significant differences in frequencies between cases and controls for any SNP analyzed. In our study, in addition to genotype frequencies, allele frequencies were determined for cases and controls, and for three SNP located in BRAF, significant differences in frequencies between cases and controls were found, suggesting BRAF is a melanoma predisposition gene. A similar conclusion was reached byMeyer et al., 2003bMeyer P. Sergi C. Garbe C. Polymorphisms of the BRAF gene predispose males to malignant melanoma.J Carcinogen. 2003; 2: 7Crossref PubMed Scopus (39) Google Scholar, who found six SNP in BRAF were associated with melanoma risk in male cases (n=470), although not female cases (n=530), or males and females combined. The SNP B11 (rs1639679) and B5 (rs1267601) were associated with melanoma susceptibility in both the Australian and German case–control populations (see Table I). The number of haplotypes spanning the 175 kb BRAF locus is surprisingly limited. In passing, the observed haplotypes exhibited what has been described as a “yin–yang” phenomenon (Zhang et al., 2003Zhang J. Rowe W.L. Clark A.G. Buetow K.H. Mismatching SNP haplotype pairs observed to be common across human populations.Am J Hum Genet. 2003; 73: 1073-1081Abstract Full Text Full Text PDF PubMed Scopus (84) Google Scholar), in that the “C” haplotype differs at every site from the most common “A” haplotype. The causal variant(s) associated with melanoma risk is as yet unknown. As noted earlier, given the substantial sequencing of all exons and exon–intron junctions (Laud et al., 2003Laud K. Kannengiesser C. Avril M.F. et al.BRAF as a melanoma susceptibility candidate gene?.Cancer Res. 2003; 63: 3061-3065PubMed Google Scholar; Casula et al., 2004Casula M. Colombino M. Satta M.P. et al.BRAF gene is somatically mutated but does not make a major contribution to malignant melanoma susceptibility: The Italian melanoma intergroup study.J Clin Oncol. 2004; 22: 286-292Crossref PubMed Scopus (47) Google Scholar) it seems unlikely that there are any undiscovered coding changes common enough to account for our results. Newly described SNP in the promoter region are also unlikely to be relevant (Jackson et al., 2005Jackson S. Harland M. Turner F. et al.No evidence for BRAF as a melanoma/nevus susceptibility gene.Cancer Epidemiol Biomarkers Prev. 2005; 14: 913-918Crossref PubMed Scopus (19) Google Scholar). The hitherto best known melanoma genetic risk factor, CDKN2A, accounts for about 25% of familial melanoma cases (Pollock and Trent, 2000Pollock P.M. Trent J.M. The genetics of cutaneous melanoma.Clin Lab Med. 2000; 20: 667-690PubMed Google Scholar), but less than 0.2% of the total melanoma burden (Aitken et al., 1999Aitken J.F. Welch J. Duffy D.L. Milligan A. Green A. Martin N. Hayward N. CDKN2A variants in a population-based sample of Queensland families with melanoma.J Natl Cancer Inst. 1999; 91: 446-452Crossref PubMed Scopus (171) Google Scholar; Tsao et al., 2000Tsao H. Zhang X. Kwitkiwski K. Finkelstein D.M. Sober A.J. Haluska F.G. Low prevalence of germline CDKN2A and CDK4 mutations in patients with early-onset melanoma.Arch Dermatol. 2000; 136: 1118-1122Crossref PubMed Scopus (72) Google Scholar). Based on the observed genotype frequencies, we estimate that BRAF could account for a proportion attributable risk to develop melanoma of 1.6% in the Australian population. Our results suggest that, in addition to the high somatic mutation rate of BRAF in melanomas and nevi, germline polymorphisms in this gene also predispose to melanoma, although not to the development of nevi or freckles. One would expect that if germline BRAF mutations are associated with increased risk of melanoma, then a similar strength of relationship with cutaneous nevus count would also be observed. We did not observe such a relationship with heterozygotes carrying the “C” haplotype, and there were only eight C/C homozygotes in the entire dataset, yielding little power, suggesting that there may be multiple independent pathways to the different phenotypes. We were not surprised by the lack of association between BRAF genotype and freckling, since an overlap in the mechanisms giving rise to freckling and to melanoma is less likely. We studied an Australian case-control sample made up of 755 melanoma cases from 740 families participating in the QFMP (Aitken et al., 1994Aitken J.F. Duffy D.L. Green A. Youl P. MacLennan R. Martin N.G. Heterogeneity of melanoma risk in families of melanoma patients.Am J Epidemiol. 1994; 140: 961-963Crossref PubMed Scopus (55) Google Scholar; Aitken et al., 1996Aitken J.F. Green A. MacLennan R. Youl P. Martin N.G. The Queensland Familial Melanoma Project: Study design and characteristics of participants.Melanoma Res. 1996; 6: 155-165Crossref PubMed Scopus (47) Google Scholar; Palmer et al., 2000Palmer J.S. Duffy D.L. Box N.F. et al.Melanocortin-1 receptor polymorphisms and risk of melanoma: Is the association explained solely by pigmentation phenotype?.Am J Hum Genet. 2000; 66: 176-186Abstract Full Text Full Text PDF PubMed Scopus (436) Google Scholar), and genotyped controls were 2239 individuals without melanoma from 635 twin families (476 DZ, 159 MZ) enrolled in the BTNS (Zhu et al., 1999Zhu G. Duffy D.L. Eldridge A. et al.A major quantitative-trait locus for mole density is linked to the familial melanoma gene CDKN2A: A maximum-likelihood combined linkage and association analysis in twins and their sibs.Am J Hum Genet. 1999; 65: 483-492Abstract Full Text Full Text PDF PubMed Scopus (192) Google Scholar,Zhu et al., 2004Zhu G. Evans D.M. Duffy D.L. et al.A genome scan for eye color in 502 twin families: Most variation is due to a QTL on chromosome 15q.Twin Res. 2004; 7: 197-210Crossref PubMed Scopus (108) Google Scholar). For purposes of a case–control analysis of melanoma, we used a subsample comprising 720 genotyped melanoma cases (one per family) and 1170 unrelated genotyped controls (parents of the twins individuals) used in calculations as indicated below. Ancestry of the cases and controls was recorded (grandparental country of birth and ethnicity), along with phenotypic risk factors such as hair and eye color, and tanning type. Tumor thickness and level were recorded for the cases that included both in situ and invasive melanomas. A familial risk index was generated for the cases using a permutation-based procedure as described elsewhere (Aitken et al., 1994Aitken J.F. Duffy D.L. Green A. Youl P. MacLennan R. Martin N.G. Heterogeneity of melanoma risk in families of melanoma patients.Am J Epidemiol. 1994; 140: 961-963Crossref PubMed Scopus (55) Google Scholar). From this, three familial risk levels (“low” (or “sporadic”), “intermediate,” and “high”) were classified that correspond approximately to having zero, one, or two or more first degree relatives affected with melanoma. These were used to further stratify the cases in terms of genetic risk. Only clinically verified cases among relatives were included in the analysis. Standard melanoma risk factors, including propensity to burn in the sun, pigmentation (skin color, hair color at 21 y, eye color, total freckling in summer, and density of melanocytic nevi) were obtained by mailed questionnaire with intensive telephone follow-up. Hair color was recorded as either black, dark brown, light brown, fair, or red/auburn and skin color was recorded as fair, medium, or olive/dark. A three-point scale was also used for eye color, including the categories blue/gray, green/hazel, and brown. Total freckling in summer was self-reported as either nil, 1–100, or 101+, and density of melanocytic nevi was estimated using a scale of nil, few, moderate, or many. Skin color (on a three-point scale), hair color (on a five-point scale), and eye color (on a three-point scale) is therefore available for most of the cases and controls. Although nevus counts were only carried out in the adolescents, self-assessed nevus number on a four-point scale (“none,”“few,”“moderate,”“many”) is available for parents of the adolescents, as well as the melanoma cases. Additional analyses of mole count has been performed using the BTNS twins and their siblings closest in age who have all undergone total body skin examination at age 12 y by a trained nurse who assessed nevus count, degree of freckling, and pigmentary phenotypes (Zhu et al., 1999Zhu G. Duffy D.L. Eldridge A. et al.A major quantitative-trait locus for mole density is linked to the familial melanoma gene CDKN2A: A maximum-likelihood combined linkage and association analysis in twins and their sibs.Am J Hum Genet. 1999; 65: 483-492Abstract Full Text Full Text PDF PubMed Scopus (192) Google Scholar). Parents of the twins were genotyped, but only self-reported pigmentary characteristics are available, employing identical measures to those used to record these characteristics in the melanoma individuals. A four-point scale, however, was used to record freckle density, including no freckling, a few, some, and many freckles. No members of the BTNS had been diagnosed with melanoma at the time of this study. Approval to undertake this study was granted by the Human Research Ethics Committee of the Queensland Institute of Medical Research. All participants gave their signed informed consent. Australian National Health and Medical Research Council encompassing the Declaration of Helsinki Guidelines for human research were adhered to. Fifteen intronic/promoter and 1 exonic BRAF SNP were typed chosen on the basis of polymorphism in our collection and distribution across the gene (Table I, Figure 1). In order to define flanking areas where linkage disequilibrium decayed around the 175 kb BRAF gene region we also typed five SNP in three flanking genes (MRPS33, ADCK2, and Q9ULE3), which extended an additional 232 and 77 kb on either side and spanning 484 kb in total. SNP identity and type is given in Table I; full sequence and other linked information may be found in the public databases by using the unique “rs” accession number (National Center for Biotechnology Information (NCBI), 2003National Center for Biotechnology Information (NCBI) Single nucleotide polymorphism database.Build. 2003; 118http://www.ncbi.nlm.nih.gov/dbSNPGoogle Scholar). Genotyping was performed via a primer extension reaction and matrix-assisted laser desorption ionization-time of flight mass spectrometry (MassARRAY, Sequenom, San Diego, California) as previously described (Bansal et al., 2002Bansal A. van den Boom D. Kammerer S. et al.Association testing by DNA pooling: An effective initial screen.Proc Natl Acad Sci USA. 2002; 99: 16971-16974Crossref Scopus (138) Google Scholar; James et al., 2004James M.R. Hayward N.K. Dumenil T. Montgomery G.W. Martin N.G. Duffy D.L. EGF polymorphism and risk of melanocytic neoplasia.J Invest Dermatol. 2004; 123: 760-762Abstract Full Text Full Text PDF PubMed Scopus (41) Google Scholar). All SNP had dropout rates of <0.5%. Data cleaning involved examining Mendelian inconsistencies, departures from Hardy–Weinberg Equilibrium, and discordances between MZ pairs. After examination, where the Mendelian errors were encountered the entire family was dropped from analysis. Error rates due to genotyping technical causes, estimated by replicate typing and use of 159 MZ twin pairs, were found to be 0.11%–0.2%, respectively. The use of the BTNS families as controls greatly increases our ability to detect genotyping problems, but makes the analysis more complex. We have extracted unrelated cases and controls, and performed multiple logistic regression analysis versus individual SNP, and Fisher–Irwin Exact tests comparing genotype counts in cases and controls. Correction for multiple testing versus individual SNP was done by repeatedly permuting cases and controls, retaining the best text statistic out of 21 tests from each of 100,000 replicates. Given that this is a replication of a previously reported association it is not necessary to apply a genome-wide level correction. Haplotypic association analysis was performed using the haplo.stat package (Lake et al., 2002Lake S. Lyon H. Silverman E. Weiss S. Laird N. Schaid D. Estimation and tests of haplotype-environment interaction when linkage phase is ambiguous.Hum Heredity. 2002; 55: 56-65Crossref Scopus (391) Google Scholar) running in R (R Development Core Team (RDCG), 2004R Development Core Team (RDCG) R: A Language and Environment for Statistical Computing.3-90005100-3 R Foundation for Statistical Computing, Vienna, Austria2004Google Scholar), cross-checking with the COCAPHASE program (Lang et al., 2003Lang J. Boxer M. MacKie R. Absence of exon 15 BRAF germline mutations in familiar melanoma.Hum Mutat. 2003; 21: 327-330Crossref PubMed Scopus (58) Google Scholar). These programs use an EM algorithm to enumerate all legal haplotypes that could give rise to an observed genotype and estimate the posterior probabilities for these haplotypes for each individual. Association is then assessed via binomial (or ordinal logistic or linear) regression using the haplotype probabilities as weights. This approach allows covariates to be easily included. Proportion attributable risk was calculated using the control genotype frequencies and the estimated genotypic odds ratios (PAR=1-1/(p2g2+2p(1-p)g1+(1-p)2), where p is the risk allele frequency, and g2 and g1 the genotypic relative risks for risk allele homozygotes and heterozygotes). NCBI-Genbank has renumbered Braf codons such that the previous 599 is now 600. Supported by grants from the Australian NHMRC (961061, 981339, 199600), the Queensland Cancer Fund, the CRC for Discovery of Genes for Common Diseases, and the US National Cancer Institute (CA88363). We thank Ann Eldridge, Marlene Grace, Megan Campbell, and Anjali Egan for technical assistance, and the melanoma patients, twins, and their families for their cooperation. Authors' disclosures of potential conflicts of interest: N. G. M. is a former member of the Scientific Advisory Board of Sequenom Inc. (he is no longer on the SAB), and holds share options in Sequenom Inc.