Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cancer Epidemiol Biomarkers Prev. Author manuscript; available in PMC 2016 July 1.
Published in final edited form as:
PMCID: PMC4573541

Functional Variants in Notch Pathway Genes NCOR2, NCSTN, and MAML2 Predict Survival of Patients with Cutaneous Melanoma



The Notch signaling pathway is constitutively activated in human cutaneous melanoma to promote growth and aggressive metastatic potential of primary melanoma cells. Therefore, genetic variants in Notch pathway genes may affect the prognosis of cutaneous melanoma patients.


We identified 6,256 SNPs in 48 Notch genes in 858 cutaneous melanoma patients included in a previously published cutaneous melanoma genome-wide association study dataset. Multivariate and stepwise Cox proportional hazards regression and false-positive report probability corrections were performed to evaluate associations between putative functional SNPs and cutaneous melanoma disease-specific survival. Receiver operating characteristic curve was constructed, and area under the curve was used to assess the classification performance of the model.


Four putative functional SNPs of Notch pathway genes had independent and joint predictive roles in survival of cutaneous melanoma patients. The most significant variant was NCOR2 rs2342924 T>C (adjusted HR, 2.71; 95% confidence interval, 1.73–4.23; Ptrend = 9.62 × 10−7), followed by NCSTN rs1124379 G>A, NCOR2 rs10846684 G>A, and MAML2 rs7953425 G>A (Ptrend = 0.005, 0.005, and 0.013, respectively). The receiver operating characteristic analysis revealed that area under the curve was significantly increased after adding the combined unfavorable genotype score to the model containing the known clinicopathologic factors.


Our results suggest that SNPs in Notch pathway genes may be predictors of cutaneous melanoma disease-specific survival.


Our discovery offers a translational potential for using genetic variants in Notch pathway genes as a genotype score of biomarkers for developing an improved prognostic assessment and personalized management of cutaneous melanoma patients.


Genetic variants, such as SNPs, have been associated with individual variation in susceptibility to cancer and in outcome of cancer treatment (1, 2). There are several genome-wide association studies (GWASs) that have identified a few SNPs associated with risk of cutaneous melanoma (37). This GWAS approach has also been used for identifying SNPs predicting survival of cutaneous melanoma patients (810). Considering the diversity of genetic and epigenetic factors involved in the origin and progress of cutaneous melanoma (11), it is very likely that SNPs in other developmental and oncogenic pathways may contribute to the variation in treatment outcomes of cutaneous melanoma patients and thus affect the survival of cutaneous melanoma patients.

The Notch signaling pathway is evolutionarily conserved in most multicellular organisms, involving gene regulation mechanisms that control cell fate determination, cell differentiation, cell proliferation, apoptosis, and cell death. A series of studies have shown that the Notch signaling plays vital roles in maintaining immature status of the melanoblast, controlling proper location of the melanoblast, and preventing migration of differentiated melanocytes to ectopic locations outside the hair matrix (12). Reports also demonstrated that the Notch pathway was activated in melanoma and that suppression of the Notch pathway could inhibit melanoma growth (13). More importantly, a gradually elevated expression pattern of the Notch signals was observed from nevi, primary melanoma to metastatic melanoma (14, 15).

Despite evidence that Notch signaling is dysregulated in many malignant tumors, including T-cell acute lymphoblastic leukemia (T-ALL) and cancers of the breast, lung, prostate, and skin (16), there are few published studies that have investigated the roles of genetic variants in Notch pathway genes in the etiology of cutaneous melanoma (17). Moreover, none of the published studies has investigated the prognostic role of genetic variants of the Notch pathway genes in cutaneous melanoma patients. Thus, we took a pathway-based multigene approach to identify putatively functional SNPs in genes involved in the Notch pathway and examined their associations with survival of cutaneous melanoma patients by using the available genotyping data from a previously published GWAS study of cutaneous melanoma (4).

Materials and Methods

Study populations

Participant recruitment and patients' characteristics have been described elsewhere (4). In brief, newly diagnosed cutaneous melanoma patients were consecutively recruited from The University of Texas M.D. Anderson Cancer Center between October 1999 and October 2007. All cases were diagnosed with histologically confirmed cutaneous melanoma, and there were no age, sex, or stage restrictions. Among the 1,804 patients, 943 patients were excluded from the analysis because of no questionnaire data. Three additional patients were excluded due to loss to the follow-up after diagnosis. Hence, the final analysis included 858 non-Hispanic white patients who had complete information about both questionnaire and clinical prognostic variables. The age of patients was between 17 and 94 years at diagnosis (52.4 ± 14.4 years). There were more stage I/II patients (709, 82.6%) than stage III/IV patients (149, 17.4%). The patients had a median follow-up time of 81.1 months, during which 95 (11.1%) died of cutaneous melanoma at the last follow-up (9). All patients provided a written informed consent under an Institutional Review Board–approved protocol.

SNP genotyping

The genotype data in the present study can be accessed by using the National Center for Biotechnology Information (NCBI) Database of Genotypes and Phenotypes (dbGaP;, with the study accession number phs000187.v1.p1. The detailed genotyping information and data quality control have been reported (4). Genome-wide imputation was performed using the MACH software based on the 1000 Genomes project (, phase I V2 CEU data (18).

SNP selection for Notch pathway analysis

Based on the databases of Kyoto Encyclopedia of Genes and Genomes (KEGG;, 48 genes located on the autosomes for the Notch signaling pathway were selected. As a result, 6,256 (955 genotyped and 5,301 imputed) SNPs within these genes or in their ±2-kb flanking regions were selected for association analyses. After quality control [i.e., minor allele frequency (MAF) ≥ 0.05, genotyping rate ≥ 95%, Hardy–Weinberg equilibrium P value ≥ 0.01, and imputation r2 ≥ 0.8], 4,949 common SNPs (902 genotyped and 4,047 imputed) in the Notch pathway genes were extracted from the cutaneous melanoma GWAS dataset. For the illustrative purpose, a flow chart of detailed SNP selection among Notch pathway genes is shown in Supplementary Fig. S1.

False-positive report probability

False-positive report probability (FPRP) is the probability of no true association between a genetic variant and disease given a statistically significant finding (19). It depends on three factors: the assumed prior probability of a true association of the tested genetic variant with a disease, an observed P value, and statistical power to detect the OR of the alternative hypothesis at the given P value. For the results of all the selected SNPs, we assigned a prior probability of 0.1 to detect an HR of 2.0 for an association with genotypes and alleles of each SNP. Only the results with an FPRP value < 0.2 were considered significant.

Statistical methods

Cutaneous melanoma disease-specific survival (DSS) served as a prognostic value was evaluated in the present study. The DSS time was calculated from the date of diagnosis to the date of death from cutaneous melanoma or date of the last follow-up, and individuals who died of causes other than cutaneous melanoma were considered censored. Associations between SNPs and DSS were obtained by multivariable Cox proportional hazards regression models performed with the GenABEL package of R software (first in an additive genetic model; ref. 20) with adjustment for age, sex, tumor stage, Clark level, Breslow thickness, ulceration of tumor, sentinel lymph node biopsy (SLNB), and tumor cell mitotic rate, which were significant predictors in the univariate Cox models for DSS. The FPRP cutoff of 0.2 was applied to limit the possibility of false-positive findings because of a relatively large number of SNPs being tested. Then, the significant SNPs were included together with clinical prognostic variables into a multivariable, stepwise Cox model. Linkage disequilibrium (LD) analysis was performed by Haploview 4.2 software to measure the degree to which alleles at two loci are associated. Breslow thickness, SLNB, tumor ulceration, andmitotic rate are required for staging melanoma patients using the seventh edition of the American Joint Committee on Cancer (AJCC) melanoma staging system (21), and these clinicopathologic factors help determine the stage of melanoma patients (but not vice versa). As a result, we also assessed the SNP-survival associations with adjustment of age, sex, and stage only to compare the differences. Because the tagging SNPs used in the GWAS chip are likely not to have some true association signals, we focused on those truly potential functional SNPs in the final analysis. To this end, the online tool RegulomDB ( was used to predict putative functions of the selected SNPs (22), by which SNPs with a score lower than 5 were considered functional. The number of unfavorable genotypes of SNPs with putative functions that were identified from the stepwise Cox models for DSS were combined as a genotype score (under a dominant genetic model) for further analyses. Kaplan–Meier survival curves and log-rank tests were used to evaluate the effects of genetic variants on the cumulative probability of DSS and overall survival (OS). We also explored the role of unfavorable genotypes in stratified analyses by age, sex, tumor stage, Clark level, Breslow tumor thickness, ulceration of tumor, SLNB, and tumor cell mitotic rate. The heterogeneity among subgroups was assessed with the χ2-based Q test, and the test was considered significant when P < 0.10. Receiver operating characteristic (ROC) curve was illustrated with the estimates obtained from the logistic regression model, and the area under the curve (AUC) was used to assess the classification performance of the model. Statistical significance of the improvement in AUC after adding an explanatory factor was calculated and evaluated by the Delong test (23). To provide biologic context for the findings, linear regression analysis was also used to test for the trends in the associations between the number of minor allele of SNPs and corresponding gene expression levels from the 270 lymphoblastoid cell lines derived from diverse populations (publicly available from the HapMap website: All other analyses were performed using SAS software (Version 9.3; SAS Institute).


Multivariate analyses of associations between SNPs and cutaneous melanoma DSS

We first performed multivariate Cox models to assess the associations of 4,949 SNPs (Supplementary Table S1) of the Notch pathway genes with DSS in the presence of age, sex, tumor stage, Breslow thickness, SLNB, Clark level, ulceration of tumor, and tumor cell mitotic rate. The results showed that 181 SNPs were individually and significantly associated with DSS at P < 0.05 in an additive genetic model (Supplementary Fig. S2), and 78 of these 181 SNPs were still considered noteworthy after the correction by FPRP (Supplementary Table S2). These 78 SNPs were all included together with clinical prognostic variables in a multivariable stepwise Cox model, in which 13 SNPs (Supplementary Table S3) remained significantly associated with DSS at P < 0.05.

Functional variants in the Notch pathway genes as independent cutaneous melanoma survival predictors

Among the 13 SNPs (Supplementary Table S3), there were two SNPs in NCOR2, six SNPs in MAML2, and other five SNPs in five other genes. When we applied the 13 significant SNPs in RegulomeDB, four were predicted to be putatively functional, including two NOCR2 SNPs (rs2342924 T>C and rs10846684 G>A), one NCSTN SNP (rs1124379 G>A), and one MAML2 SNP (rs79453425 G>A). We then performed LD analysis on NCOR2 and MAML2 because there were more than one significant SNP in these two genes. As shown in Supplementary Fig. S3, there were low LD between the two SNPs in NCOR2 (r2 = 0.07) and low LD among the six SNPs in MAML2 (r2 values range from 0 to 0.12). These four putatively functional SNPs were also analyzed for their roles in predicting DSS and OS in the presence of other clinicopathologic covariates in multivariate Cox models (Table 1; Supplementary Table S4). The associations of NCOR2 rs2342924C and rs10846684A, NCSTN rs1124379A, and MAML2 rs79453425A with DSS were statistically significant in a trend test (P = 9.62E−07, 0.005, 0.005, and 0.013, respectively; Table 1). Compared with their homozygous genotypes, these unfavorable (variant) genotypes in a dominant genetic model were significantly associated with a poor DSS [HR, 2.71, 95% confidence interval (95% CI), 1.73–4.23, and P = 1.28E−05 for rs2342924 CC+CT; 1.64, 1.07–2.51, and 0.022 for rs10846684 AA+AG; 2.36, 1.28–4.36 and 0.006 for rs1124379 AG+GG; and 1.77, 1.09–2.89, and 0.021 for rs79453425 AA+AG; Table 1]. Similar results were obtained when performing multivariate analyses with adjustment only for age, sex, and tumor stage (data not shown). These four SNPs were also significantly associated with OS, though there were some changes in the HR and P values (Supplementary Table S4). The regional association results from the GWAS dataset were plotted for these three genes (with 2-kb flanking the neighborhood of NCOR2, NCSTN, and MAML2; Supplementary Fig. S4).

Table 1
Association between potential SNPs in the Notch pathway genes and DSS of cutaneous melanoma patients

Cutaneous melanoma DSS predicted by the combined unfavorable genotypes of the four SNPs

To better estimate the joint effect of the four SNPs on patients' clinic outcomes, we assessed the DSS associated with the combined unfavorable genotypes (a genotype score under a dominant genetic model) of rs2342924 CC+CT, rs10846684 AA+AG, rs1124379 AG+GG (this was under a recessive model), and rs79453425 AA+AG. The frequencies of 0, 1, 2, 3, and 4 of the unfavorable genotype score were 51, 264, 367, 160, and 16, respectively. For the illustrative purpose, Kaplan–Meier survival curves of the associations of DSS and OS with the unfavorable genotype score are shown in Fig. 1A and B. In the multivariate Cox models, the per-unit increase of unfavorable genotype score was statistically significantly associated with a poor DSS (Ptrend = 3.48E−10) in a trend test with adjustment for age, sex, tumor stage, Breslow thickness, SLNB, Clark level, ulceration of tumor, and tumor cell mitotic rate (Table 1). A similar trend in the associations was observed between melanoma OS and the combined unfavorable genotype score (Ptrend = 5.4E−10; Supplementary Table S4).

Figure 1
Kaplan–Meier (KM) estimates of melanoma survival by unfavorable genotype numbers. KM estimates of melanoma-specific survival by the exact numbers of unfavorable genotypes (A) and the dichotomized numbers of unfavorable genotypes (C); OS function ...

To provide a larger and stable reference group, we then divided the combined unfavorable genotype score into two groups: low-risk group (0–1) and high-risk group (2–4). Kaplan–Meier survival curves of the associations of DSS and OS in cutaneous melanoma patients with 0–1 and 2–4 unfavorable genotype score are shown in Fig. 1C and D, respectively. In the multivariate Cox models, compared with the low-risk group, both DSS and OS were reduced significantly in the high-risk group [adjusted hazard ratio (adjHR) = 3.98, 95% CI, 2.26–6.99, P = 1.68E−06 for DSS (Table 1) and adjHR = 3.19, 95% CI, 2.03–5.02, P = 4.71E−07 for OS (Supplementary Table S4)].

Stratified analyses for unfavorable genotype score and cutaneous melanoma DSS

To investigate whether the combined effect of unfavorable genotype score on cutaneous melanoma survival was modified by some important clinicopathologic factors, we performed stratified analyses. To better illustrate the differences between cutaneous melanoma patients with 0–1 and 2–4 unfavorable genotype score, Kaplan–Meier curves of DSS were plotted by tumor-related characters (Fig. 2). As shown in Table 2 and Supplementary Table S5, compared with those with the score of 0–1, those with a score of 2–4 had significantly decreased survival rate in the presence or absence of clinicopathologic risk factors in most of stratified subgroups, except for the subgroups of Clark level II/III, Breslow thickness ≤ 1.0 mm, and mitotic rate < 1 mitoses/mm2. Notably, the adjHR for DSS associated with 2–4 unfavorable genotype score, compared with 0–1 unfavorable genotype score, was 2.10 (1.06–4.14) for stage I/II patients but 9.99 (3.40–29.3) for stage III/IV patients and, similarly, 2.16 (1.09–4.25) for patients with negative SLNB but 9.91 (3.38–29.1) for patients with positive SLNB. However, these differences by subgroup were not statistically different by the heterogeneity test, likely due to small numbers in the subgroups.

Figure 2
Kaplan–Meier (KM) estimates of melanoma-specific survival by dichotomized unfavorable genotypes for patients with age ≤ 50 (A), age > 50 (B); male (C) and female (D); stage I/II (E) and III/IV (F); Clark level II/III (G) and IV/V ...
Table 2
Stratified association analyses on DSS and HRs for cutaneous melanoma patients with different numbers of risk genotypes across genes in the Notch pathway

The ROC curve

Using multivariate logistic regression and ROC curve, we further evaluated the unfavorable genotype score for their potential to improve the classification of 5-year DSS and OS. As shown in Fig. 3, the AUC of the 5-year DSS and OS models significantly increased from 82.0% and 74.7%, respectively, with clinical variables as classifiers alone, to 85.2% and 78.2%, respectively, with these classifiers plus the risk genotypes (P = 0.008 and P = 0.001, respectively, as assessed by the DeLong test). These results suggest a potential role of the unfavorable genotype score in predicting cutaneous melanoma DSS and OS.

Figure 3
ROC curves for prediction of 5-year melanoma-specific survival rate (A) and overall survival rate (B) based on only clinical variables (tumor stage, Breslow tumor thickness, Clark level, and ulceration of tumor) and combined risk genotypes along with ...

Genotype–phenotype correlation analyses

Finally, we used the publicly available expression data of the HapMap 270 normal lymphoblastoid cell lines to further evaluate the correlations between SNPs and their corresponding gene mRNA expression levels. Such expression data are available for NCOR2 rs2342924 and rs10846684, and NCSTN rs1124379 but not for MAML2 rs79453425. As shown in Fig. 4, the rs2342924C allele was associated with significantly lower levels of mRNA expression of NCOR2 (P = 0.044), but such a genotype–phenotype correlation was not evident for rs10846684 and rs1124379.

Figure 4
Analyses of corresponding gene expression levels by genotypes of NCOR2 rs2342924 (A), rs10846684 (B), and NCSTN rs1124379 (C) using 270 HapMap lymphoblastoid cell lines of all population. Genotypes CT/CC of SNP rs2342924 were significantly associated ...


In the present study, we comprehensively investigated the predictive role of putatively functional variants in the Notch pathway genes in cutaneous melanoma DSS using the published GWAS dataset. We found that NCOR2 rs2342924 T>C, rs10846684 G>A, NCSTN rs1124379 G>A, and MAML2 rs79453425 G>A independently or jointly modulated survival of cutaneous melanoma patients. Our results suggest that Notch pathway genes may have a biologic implication in cutaneous melanoma progression.

There is evidence suggesting that Notch pathway genes involved in tumorigenesis (16, 24, 25). This pathway may act either as a tumor promoter or suppressor, depending on the cell type and tissue context, levels of expression and potential crosstalk with other signaling pathways (26). In humans, the constitutively activated Notch signaling enhances growth and aggressive metastatic potential of primary melanoma cells both in vitro and in vivo (27). However, no study has reported a role of genetic variants of Notch pathway genes in predicting clinical outcomes of cancer.

In the present study, three putatively functional SNPs of Notch coregulators were found to be significantly associated with cutaneous melanoma DSS and OS. Specifically, carriers of the NOCR2 rs2342924T and rs10846684G and MAML2 rs79453425G variant genotypes had a better DSS, compared with those with CC, AA, and AA homozygous genotypes, respectively. Among these three SNPs, rs10846684 and rs2342924 are located at the first and third introns of NCOR2, respectively, whereas rs79453425 is located at the second intron of MAML2. The online prediction tool RegulomeDB for analysis of DNase-seq showed that rs2342924, rs10846684, and rs79453425 are located in the DNase I hypersensitive sites, which represent open and active chromatins. Additional ChIP-seq data indicated that these variants were located in the enhancer region containing histone modification marks of H3k4me1 and H3k27ac. Thus, these three SNPs are likely to affect the binding of transcriptional factors and thus to modify the function of regulatory elements.

By searching a published expression data containing 270 HapMap of lymphoblastoid cell lines derived from diverse populations (28), we found that the unfavorable CC+TT genotypes of rs2342924 were shown to be associated with lower mRNA expression levels of NCOR2. This genotype–phenotype correlation also provides additional biologic evidence that NCOR2 expression may be mediated by this putatively functional SNP rs2342924, a possible explanation for the observed association with cutaneous melanoma DSS. NCOR2, also known as a silencing mediator for retinoid or thyroid-hormone receptors (SMRT), is a Notch pathway corepressor and located at 12q24. Although the precise role of NCOR2 in carcinogenesis remains uncertain, it was observed that the elevated nuclear expression of NCOR2 was correlated with poor outcomes in breast cancer patients and with earlier tumor recurrence in breast cancer patients not receiving adjuvant tamoxifen therapy (29, 30). Mechanistic studies have shown that recruitment of NCOR2 can downregulate the IL6-mediated cancer cell growth and gene expression by transcriptionally inactivating STAT3 (31), whereas silencing NCOR2 could lead to cell circle progression (32).

MAML2 encodes another Notch pathway coregulator that was found to be associated with cutaneous melanoma DSS in our analysis. MAML2 is located at 11q21, and its encoding protein is capable of forming a multiprotein complex with NIC-RBPJκ, which is an essential step for the Notch-mediated transcriptional activation (33). The oncogenic role of MAML2 was first described in mucoepidermoid carcinoma, in which translocation of MAML2 in mucoepidermoid carcinoma will create a fusion oncogene mucoepidermoid carcinoma translocated 1 (MECT1)—MAML2 that is involved in disrupting the normal cell cycle, differentiation, and tumor development (34). Clinical investigation also demonstrated that mucoepidermoid carcinoma patients with a positive MECT1–MAML2 fusion and MAML2 gene split had significantly longer OS (34, 35). It was reported that MECT1–MAML2 could bind to and activate both c-jun and c-fos, which are known as proto-oncogenes (36). A gain-of-function study also showed that MECT1–MAML2 could activate oncogene MYC and in turn activate MYC transcription targets, including those involved in cell growth and metabolism, survival, and tumorigenesis (37). These studies provided some biologic evidence for the role played by MAML2 in possible molecular mechanisms underlying our observed associations.

The other SNP associated with DSS of cutaneous melanoma patients in the Notch pathway was NCSTN rs1124379, located in intron 7 of the gene. Carriers of rs1124379 A variant allele had a better DSS compared with those GG homozygotes in cutaneous melanoma patients. ChIP-seq data on RegulomeDB suggested that rs1124379 may influence the binding activity of transcriptional factor RFX5, as the SNP is located in its binding sites. NCSTN, also referred to as nicastrin, is located at 1q22–q23 and encodes a type I transmembrane glycoprotein that is one of four core subunits of the γ-secretase complex. NCSTN is a stabilizing cofactor required for the γ-secretase complex assembly and can cleave transmembrane domains of Notch receptors (25). The roles of NCSTN have been investigated in several non-melanoma cancers. For instance, NCSTN functions to maintain epithelial to mesenchymal transition during breast cancer progression, and its high expression can be used as a predictor for worse breast cancer–specific survival in the ERα-negative cohort (38); Others reported that NCSTN overexpression was detected in both cell lines and clinical sample of T-ALL (39) and that a monoclonal antibody of NCSTN, which could recognize extracellular domain of NCSTN, inhibited the γ-secretase activity and abolished the γ-secretase activity–dependent growth of cancer cells (40). Thus, targeting NCSTN might be a new therapeutic strategy. Further functional studies of the gene in cutaneous melanoma are warranted to provide biologic support for this observed association.

In the present study, we found that the combined numbers of unfavorable genotypes of the four Notch pathway SNPs could improve prediction of cutaneous melanoma patients' survival; that is, a reduced survival was associated with an increasing number of unfavorable genotype score. The results were in line with the concept that a pathway-based multigene approach could magnify the effects of individual variant or gene to have a better prediction of the prognosis, compared with analyses of each single variant or gene. The effect was consistent across different analyses and multiple subgroup comparisons, regardless of other clinicopathologic characteristics. In the presence of the previously verified clinicopathologic prognostic characteristics of the melanoma patients, such as tumor stage, Clark level, Breslow tumor thickness, and ulceration of tumor [21], the combination of unfavorable genotypes, as shown in the ROC analysis, significantly improved the predictive power of DSS and OS.

In fact, through stratified analyses, we found that the genotype–survival association was more pronounced in the presence of clinicopathologic risk factors, such as late tumor stage, presence of ulceration and positive SLNB. These results suggest that these SNPs in the Notch pathway may aggregate the existing genomic instability of highly malignant melanoma, promoting melanoma development, and progression in the high-risk populations. Therefore, the present study identified a significant proportion of melanoma patients (such as those with unfavorable genotypes) that may require close clinical surveillance or alternative treatment to improve their survival.

However, there were some limitations in the present study. Firstly, we were unable to explore the exact mechanisms by which the Notch pathway SNPs influence DSS, because we did not have the access to the target tissues. Secondly, although the present study included a relatively large number of cutaneous melanoma patients, due to the limitation of available clinical data and a limited number of the events, we were unable to evaluate the potential role of the SNPs by different therapies that might provide specific survival benefit, although the vast majority of the patients had early stage of cutaneous melanoma. Thirdly, we did not find a suitable and accessible patient population for the validation of our results. Finally, additional larger validation studies with multiethnic groups are needed to confirm our results, because our prognosis-predicting model was based on a non-Hispanic white patient population.

Supplementary Material

Supplemental Figures

Supplemental Results


The authors thank the individuals who participated in this project, the Johns Hopkins University Center for Inherited Disease Research for conducting high-throughput genotyping for this study, and the University of Washington for the performance of quality control of the high-density SNP data.

Grant Support

This work was supported by the NIH, National Cancer Institute Grants R01 grant CA100264 (to Q. Wei), the National Cancer Institute MD Anderson Cancer Center SPORE in Melanoma P50 CA093459 (to E.A. Grimm and J.E. Lee), and the Marit Peterson Fund for Melanoma Research (to J.E. Lee). This work was also supported by a start-up funds (to Q. Wei) from Duke Cancer Institute, Duke University Medical Center and support from the Duke Cancer Institute as part of the P30 Cancer Center Support Grant (NIH CA014236).


Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Authors' Contributions

Conception and design: W. Zhang, H. Liu, Q. Wei

Development of methodology: W. Zhang, Q. Wei

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Liu, C.I. Amos, S. Fang, J.E. Lee, Q. Wei

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Zhang, H. Liu, Z. Liu, C.I. Amos, S. Fang, Q. Wei

Writing, review, and/or revision of the manuscript: W. Zhang, H. Liu, Z. Liu, C.I. Amos, S. Fang, J.E. Lee, Q. Wei

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Zhang, H. Liu, Z. Liu, D. Zhu, C.I. Amos, J.E. Lee, Q. Wei

Study supervision: Q. Wei


1. Webb T. SNPs: can genetic variants control cancer susceptibility? J Natl Cancer Inst. 2002;94:476–478. [PubMed]
2. Erichsen HC, Chanock SJ. SNPs in cancer research and treatment. Br J Cancer. 2004;90:747–751. [PMC free article] [PubMed]
3. Bishop DT, Demenais F, Iles MM, Harland M, Taylor JC, Corda E, et al. Genome-wide association study identifies three loci associated with melanoma risk. Nat Genet. 2009;41:920–925. [PMC free article] [PubMed]
4. Amos CI, Wang LE, Lee JE, Gershenwald JE, Chen WV, Fang S, et al. Genome-wide association study identifies novel loci predisposing to cutaneous melanoma. Hum Mol Genet. 2011;20:5012–5023. [PMC free article] [PubMed]
5. Brown KM, Macgregor S, Montgomery GW, Craig DW, Zhao ZZ, Iyadurai K, et al. Common sequence variants on 20q11.22 confer melanoma susceptibility. Nat Genet. 2008;40:838–840. [PMC free article] [PubMed]
6. Teerlink C, Farnham J, Allen-Brady K, Camp NJ, Thomas A, Leachman S, et al. A unique genome-wide association analysis in extended Utah high-risk pedigrees identifies a novel melanoma risk variant on chromosome arm 10q. Hum Genet. 2012;131:77–85. [PubMed]
7. Nan H, Xu M, Zhang J, Zhang M, Kraft P, Qureshi AA, et al. Genomewide association study identifies nidogen 1 (NID1) as a susceptibility locus to cutaneous nevi and melanoma risk. Hum Mol Genet. 2011;20:2673–2679. [PMC free article] [PubMed]
8. Rendleman J, Shang S, Dominianni C, Shields JF, Scanlon P, Adaniel C, et al. Melanoma risk loci as determinants of melanoma recurrence and survival. J Transl Med. 2013;11:279. [PMC free article] [PubMed]
9. Yin J, Liu H, Liu Z, Wang LE, Chen WV, Zhu D, et al. Genetic variants in fanconi anemia pathway genes BRCA2 and FANCA predict melanoma survival. J Invest Dermatol. 2015;135:542–550. [PMC free article] [PubMed]
10. Li C, Yin M, Wang LE, Amos CI, Zhu D, Lee JE, et al. Polymorphisms of nucleotide excision repair genes predict melanoma survival. J Invest Dermatol. 2013;133:1813–1821. [PMC free article] [PubMed]
11. Meyle KD, Guldberg P. Genetic risk factors for melanoma. Hum Genet. 2009;126:499–510. [PubMed]
12. Moriyama M, Osawa M, Mak SS, Ohtsuka T, Yamamoto N, Han H, et al. Notch signaling via Hes1 transcription factor maintains survival of melanoblasts and melanocyte stem cells. J Cell Biol. 2006;173:333–339. [PMC free article] [PubMed]
13. Panelos J, Massi D. Emerging role of Notch signaling in epidermal differentiation and skin cancer. Cancer Biol Ther. 2009;8:1986–1993. [PubMed]
14. Hoek K, Rimm DL, Williams KR, Zhao H, Ariyan S, Lin A, et al. Expression profiling reveals novel pathways in the transformation of melanocytes to melanomas. Cancer Res. 2004;64:5270–5282. [PubMed]
15. Balint K, Xiao M, Pinnix CC, Soma A, Veres I, Juhasz I, et al. Activation of Notch1 signaling is required for beta-catenin-mediated human primary melanoma progression. J Clin Invest. 2005;115:3166–3176. [PubMed]
16. Capaccione KM, Pine SR. The Notch signaling pathway as a mediator of tumor survival. Carcinogenesis. 2013;34:1420–1430. [PMC free article] [PubMed]
17. Hicks C, Pannuti A, Miele L. Associating GWAS information with the Notch signaling pathway using transcription profiling. Cancer Inform. 2011;10:93–108. [PMC free article] [PubMed]
18. Li Y, Willer CJ, Ding J, Scheet P, Abecasis GR. MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes. Genet Epidemiol. 2010;34:816–834. [PMC free article] [PubMed]
19. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N. Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst. 2004;96:434–442. [PubMed]
20. Aulchenko YS, Ripke S, Isaacs A, van Duijn CM. GenABEL: an R library for genome-wide association analysis. Bioinformatics. 2007;23:1294–1296. [PubMed]
21. Balch CM, Gershenwald JE, Soong SJ, Thompson JF, Atkins MB, Byrd DR, et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol. 2009;27:6199–6206. [PMC free article] [PubMed]
22. Boyle AP, Hong EL, Hariharan M, Cheng Y, Schaub MA, Kasowski M, et al. Annotation of functional variation in personal genomes using RegulomeDB. Genome Res. 2012;22:1790–1797. [PubMed]
23. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–845. [PubMed]
24. Sahlgren C, Gustafsson MV, Jin S, Poellinger L, Lendahl U. Notch signaling mediates hypoxia-induced tumor cell migration and invasion. Proc Natl Acad Sci U S A. 2008;105:6392–6397. [PubMed]
25. Mullendore ME, Koorstra JB, Li YM, Offerhaus GJ, Fan X, Henderson CM, et al. Ligand-dependent Notch signaling is involved in tumor initiation and tumor maintenance in pancreatic cancer. Clin Cancer Res. 2009;15:2291–2301. [PMC free article] [PubMed]
26. Panelos J, Batistatou A, Paglierani M, Zioga A, Maio V, Santi R, et al. Expression of Notch-1 and alteration of the E-cadherin/beta-catenin cell adhesion complex are observed in primary cutaneous neuroendocrine carcinoma (Merkel cell carcinoma) Mod Pathol. 2009;22:959–968. [PubMed]
27. Pinnix CC, Herlyn M. The many faces of Notch signaling in skin-derived cells. Pigment Cell Res. 2007;20:458–465. [PubMed]
28. Holm K, Melum E, Franke A, Karlsen TH. SNPexp - A web tool for calculating and visualizing correlation between HapMap genotypes and gene expression levels. BMC Bioinformatics. 2010;11:600. [PMC free article] [PubMed]
29. Green AR, Burney C, Granger CJ, Paish EC, El-Sheikh S, Rakha EA, et al. The prognostic significance of steroid receptor co-regulators in breast cancer: co-repressor NCOR2/SMRT is an independent indicator of poor outcome. Breast Cancer Res Treat. 2008;110:427–437. [PubMed]
30. Keeton EK, Brown M. Cell cycle progression stimulated by tamoxifenbound estrogen receptor-alpha and promoter-specific effects in breast cancer cells deficient in N-CoR and SMRT. Mol Endocrinol. 2005;19:1543–1554. [PubMed]
31. Wang LH, Yang XY, Zhang X, Farrar WL. Nuclear receptors as negative modulators of STAT3 in multiple myeloma. Cell Cycle (Georgetown, Tex) 2005;4:242–245. [PubMed]
32. Bhaskara S, Knutson SK, Jiang G, Chandrasekharan MB, Wilson AJ, Zheng S, et al. Hdac3 is essential for the maintenance of chromatin structure and genome stability. Cancer Cell. 2010;18:436–447. [PMC free article] [PubMed]
33. Wu L, Sun T, Kobayashi K, Gao P, Griffin JD. Identification of a family of mastermind-like transcriptional coactivators for mammalian notch receptors. Mol Cell Biol. 2002;22:7688–7700. [PMC free article] [PubMed]
34. Noda H, Okumura Y, Nakayama T, Miyabe S, Fujiyoshi Y, Hattori H, et al. Clinicopathological significance of MAML2 gene split in mucoepidermoid carcinoma. Cancer Sci. 2013;104:85–92. [PubMed]
35. Behboudi A, Enlund F, Winnes M, Andren Y, Nordkvist A, Leivo I, et al. Molecular classification of mucoepidermoid carcinomas-prognostic significance of the MECT1-MAML2 fusion oncogene. Genes Chromosomes Cancer. 2006;45:470–481. [PubMed]
36. Canettieri G, Coni S, Della Guardia M, Nocerino V, Antonucci L, Di Magno L, et al. The coactivator CRTC1 promotes cell proliferation and transformation via AP-1. Proc Natl Acad Sci U S A. 2009;106:1445–1450. [PubMed]
37. Amelio AL, Fallahi M, Schaub FX, Zhang M, Lawani MB, Alperstein AS, et al. CRTC1/MAML2 gain-of-function interactions with MYC create a gene signature predictive of cancers with CREB-MYC involvement. Proc Natl Acad Sci U S A. 2014;111:E3260–E3268. [PubMed]
38. Filipovic A, Gronau JH, Green AR, Wang J, Vallath S, Shao D, et al. Biological and clinical implications of nicastrin expression in invasive breast cancer. Breast Cancer Res Treat. 2011;125:43–53. [PubMed]
39. Paryan M, Mohammadi-Yeganeh S, Samiee SM, Soleimani M, Arefian E, Azadmanesh K, et al. Investigation of deregulated genes of Notch signaling pathway in human T cell acute lymphoblastic leukemia cell lines and clinical samples. Mol Biol Rep. 2013;40:5531–5540. [PubMed]
40. Hayashi I, Takatori S, Urano Y, Miyake Y, Takagi J, Sakata-Yanagimoto M, et al. Neutralization of the gamma-secretase activity by monoclonal antibody against extracellular domain of nicastrin. Oncogene. 2012;31:787–798. [PMC free article] [PubMed]