Genomic testing refined accuracy of melanoma risk prediction

ARLINGTON, VA. – Combining test data for melanocortin-1 receptor gene variants and common genomic variants with nongenetic screening resulted in better melanoma risk prediction, Ann Cust, Ph.D., reported at the annual meeting of the American Society of Preventive Oncology.

“We’ve shown in this study that measuring genetic factors to determine who’s at high risk for melanoma can play an important role in prediction," said Dr. Cust of the University of Sydney, Australia, in an interview. “This gives us a good model for looking at whether genetic factors can improve the way we target preventive behaviors."

In the United States, melanoma accounts for only about 2% of all skin cancers, but the most skin cancer–related deaths. Fair skin, light hair, freckling, and a family history of melanoma are known risk factors for the disease. Australia, where the study was conducted, has the highest incidence of melanoma in the world.

To date, skin cancer prevention efforts in both countries have largely relied upon mass media sun protection campaigns and identifying people at high risk based on their skin and hair pigmentation,

Dr. Cust said. “But some people don’t know they are at high risk, and so they aren’t taking precautions," she said. “Some of the genetic factors for melanoma are linked to pigmentation, but there are variations in genes involved in DNA repair and other biological pathways that occur in people whose pigmentation wouldn’t suggest they are at high risk, but they are."

For the study, Dr. Cust and her colleagues genotyped common variants in 18 different genes in a study group of 552 Australians, aged 18-39 years, all of whom had confirmed cases of invasive cutaneous melanoma, and also in a control group of 405 Australians of European ancestry without melanoma. The study was a population-based, case-control family study that assessed traditional melanoma risk factors such as hair color, moles, family history of melanoma, use of indoor tanning, and tendency to sunburn. They then performed genomewide association studies to identify the 18 specific gene regions that have common genomic variants that influence melanoma risk.

The investigators found that the area under the curve (AUC) of the predicted risk in the controls went from 0.76 (95% confidence interval, 0.73-0.79) using demographic and nongenetic factors, to 0.81 (95% CI, 0.78 -0.84) when MC1R genotype and novel common genomic variant data were added. They also found that the combined contribution to the AUC of the novel common variants was similar to that of the established common variants of MC1R and CDKN2A.