Page 315 - Personalised medicine of fluoropyrimidines using DPYD pharmacogenetics Carin Lunenburg

GWAS of Low et al. focused only on neutropenia and leukopenia as toxicity endpoint.25 In the current GWAS we chose to include all types of fluoropyrimidine-induced toxicity, as we aim to improve fluoropyrimidine treatment by reducing all types of toxicity. Compared to the GWAS of O’Donnell et al. 22, we offer a cohort with clinical patient data representative of daily clinical care, in order to identify variants which could be clinically relevant. Compared to the GWAS of Fernandez-Rodzilla et al. our cohort is much larger.24
When applying prospective DPYD genotyping, still 23% of patients treated with fluoropyrimidines experience severe toxicity.5 In order to further reduce this number, other genetic variants predictive for severe toxicity need to be identified. Options are to screen for rare variants in DPYD, investigate epigenetics, or look outside of the DPYD gene as was performed in this GWAS. The onset of severe toxicity might not only be linked to the start of fluoropyrimidines, but can be multifactorial and linked to patient characteristics or co-medication. In line of that thought, the onset of toxicity might be better predicted by multiple genetic variants. For this reason, we executed the PRS analysis. In addition, the future of genotyping is quickly evolving, with less single SNP-based assays and more assays with a panel of SNPs in genes or assays sequencing entire regions of genes, leading to future possibilities to apply a PRS in patients in clinical care. Our PRS analysis showed that a panel of 5,055 SNPs combined with clinical covariates outperforms clinical parameters alone, and can predict 96% of severe fluoropyrimidine-induced toxicity. Our PRS analysis possibly shows too optimistic results due to the pre-selection of significant SNPs into the score. Although we used cross-validation to verify the score, this step did not include the SNP selection as this would have been computationally prohibitive. We see this as an exploratory analysis that needs validation, but still suggests that low penetrance variants exist which are difficult to prove in a single-variant association test. The PRS analysis shows the possibility of future research with a multifactorial research approach. The panel of SNPs needs replication in a validation cohort and additional research is needed to be able to link the result to a dose adjustment advice in order to prevent toxicity.
In conclusion, while no genome-wide significant SNPs could be identified in our unique dataset of patients, six variants were suggestive for the onset of severe toxicity. These variants are located outside of known fluoropyrimidine-pathway genes. Using a PRS consisting of 5,055 SNPs combined with clinical variables explained 96% of toxicity in an optimistic analysis, suggesting highly polygenic nature of toxicity predisposition. This GWAS is one of the first attempts to identify variants predictive for fluoropyrimidine-induced toxicity. The identified variants and the PRS require replication in an independent cohort.
12
Genome-wide association study
313