Impact of common genomic variants on melanoma risk prediction — ASN Events
  1. Schedule
  2. Best of the Best Orals - Epidemiology
  3. Impact of common genomic variants on melanoma risk prediction

Impact of common genomic variants on melanoma risk prediction (#121)

Anne E. Cust 1 , Minh Bui 2 , Elizabeth A. Holland 3 , Chris Goumas 4 , Helen Schmid 3 , Graham Giles 2 5 , Joanne Aitken 6 , Richard Kefford 3 7 , John Hopper 2 , Graham J. Mann 3 , Mark Jenkins 2
  1. Cancer Epidemiology and Services Research, Sydney School of Public Health, The University of Sydney, Sydney, NSW, Australia
  2. Centre for Epidemiology and Biostatistics, Melbourne School of Population Health, University of Melbourne, Melbourne, VIC, Australia
  3. Westmead Institute for Cancer Research, University of Sydney at Westmead Millennium Institute and Melanoma Institute Australia, Sydney, NSW, Australia
  4. University of Sydney, Camperdown, NSW, Australia
  5. Cancer Council Victoria, Melbourne, VIC, Australia
  6. Cancer Council Qld, Brisbane, QLD, Australia
  7. Macquarie University, Sydney, NSW, Australia

Background: Genome-wide association studies have identified numerous common genomic variants associated with increased susceptibility to melanoma, but there is limited knowledge about the utility of adding them to risk prediction models for melanoma.

Aim: To evaluate the contribution of common genomic variants to melanoma risk prediction, among young Australian adults.

Methods: The sample included 552 cases with invasive cutaneous melanoma diagnosed between ages 18-39 years and 405 controls from an Australian population-based, case-control-family study. MC1R genotype was sequenced, and through a genome-wide association study we obtained genotype data for single nucleotide polymorphisms from 18 selected gene regions. Measures of discriminatory accuracy included the area under receiver operating characteristic curves (AUC) and net reclassification improvement (NRI), calculated based on predicted probabilities of melanoma from unconditional logistic regression models. We used 10-fold cross-validation and bootstrap methods to assess internal validation.

Results: Compared to a demographic model containing age, sex and city of recruitment (AUC 0.69; 95% CI 0.65-0.72), the combined contribution to the AUC of common genomic variants was the same as that contributed from traditional self-reported risk factors for melanoma (UV exposure, pigmentation phenotype, nevi, etc) – both AUCs increased to 0.77 (95% CI 0.74-0.80). An inclusive model containing demographic, genetic and non-genetic (traditional) risk factors had an AUC of 0.81 (95% CI 0.78-0.84). Inclusion of genomic variants in the multivariate model improved the quartile classification of predicted risk (NRI) by a net 17% (95% CI 9-24) compared to the non-genetic (traditional) model.

Conclusions: Our results suggest that common genomic variants could considerably improve risk prediction models for early-onset melanoma, and may have a role in primary prevention of melanoma. We are commencing pilot studies to translate these findings into potential cancer prevention strategies in general practice and in the community.

#COSAASM