Introduction: The detection of a unique genetic marker is not possible for multifactorial diseases, such as thyroid cancer (TC), where the pathological phenotype is given by the contribution of multiple genes, environmental factors and lifestyle. We found a mathematical model for inferring the risk of thyroid cancer, an example of multifactorial disease.
Methods: Genetic data represented by 184 SNPs associated to thyroid tumors were used for Bayesian clustering of 2504 individuals from the 1000 Genomes database, by STRUCTURE software. Numerical values representing the inferred genetic structure of each individual is provided in the output file of the software and were matched with environmental and lifestyle parameters associated to thyroid functions, i.e. iodine exposure and obesity, by principal component analysis (PCA). Data analysis was performed using labels such as geographic origin of individuals, population, sex and thyroid cancer incidence.
Results: We found that seven thyroid tumor-related genetic clusters are differently represented among human populations. The matching of genotype, iodine and obesity data resulted in individuals gradient distribution by thyroid cancer incidence, revealing that all these components are required to infer the disease risk. Genetic background and, to a lesser extent, environmental factors and lifestyle, are not related per se to a specific range value of cancer risk. An exception is provided by individuals from Tuscany, Italy, which deviates from the overall distribution, preserving high cancer risk independently from obesity or iodine. This could result from a peculiar genetic setting or from the exposure to environmental factors not considered in the analysis.
Discussion: We demonstrated that TC risk may be detected a priori by applying polygenic model to specific population or individuals.
Conclusion: This study provides a novel mathematical approach to infer the polygenic disease risk, as a promising diagnostic tool for personalized medicine.
20 May 2017 - 23 May 2017