Radiogenomics: using genetics to identify cancer patients at risk for development of adverse effects following radiotherapy

Abstract

Normal-tissue adverse effects following radiotherapy are common and significantly affect quality of life. These effects cannot be accounted for by dosimetric, treatment, or demographic factors alone, and evidence suggests that common genetic variants are associated with radiotherapy adverse effects. The field of radiogenomics has evolved to identify such genetic risk factors. Radiogenomics has two goals: (i) to develop an assay to predict which patients with cancer are most likely to develop radiation injuries resulting from radiotherapy, and (ii) to obtain information about the molecular pathways responsible for radiation-induced normal-tissue toxicities. This review summarizes the history of the field and current research.

Significance: A single-nucleotide polymorphism–based predictive assay could be used, along with clinical and treatment factors, to estimate the risk that a patient with cancer will develop adverse effects from radiotherapy. Such an assay could be used to personalize therapy and improve quality of life for patients with cancer.

Figure 1

A predictive assay based on genetics could be used to identify the subset of patients at increased risk of developing adverse effects. These patients could be candidates for a non-radiation treatment, a more conformal form of RT (such as IMRT or protons), or possibly a lower dose. These patients could also be treated with radio-protective agents developed on the basis of genes identified through radiogenomics studies. Patients who do not have a genetic predisposition to adverse effects could receive higher doses to increase cure rate.

Figure 2

Two commonly used study designs in radiogenomics GWAS: A) two-stage design and B) meta-analysis.

Figure 3

A) Power to detect SNPs of varying minor allele frequency (MAF) that have effect sizes of 1.5 or 2 (red line) (assuming an additive genetic model) for a radiotherapy adverse effect endpoint affecting 10% of patients. Assumes a type-I error of 5×10⁻⁸. GRR, genome relative risk. B) Improved power to detect a hypothetical locus on chromosome 8 from meta-analysis of 3 individual GWAS.

Figure 4

A multivariable predictive model including SNPs, clinical and treatment factors could be used to classify patients based on risk of developing adverse effects. The results of such a model could be used, in combination with information on tumor aggressiveness, to optimize therapy for each patient.