Page 236 - Personalised medicine of fluoropyrimidines using DPYD pharmacogenetics Carin Lunenburg

Chapter 9
Discussion
The topic of confirmatory testing in the rapidly growing field of PGx deserves attention. At this moment, there are no clear guidelines on the required confirmation practice aspects of PGx testing. Should laboratories execute a second method to confirm results, or not? The FDA is in debate on this dilemma and the current guidelines of the EMA are not very precise on the use of confirmation methods. Our supporting data show that there is great heterogeneity between laboratories in confirmation practice. Discrepant results were identified between two tests in about 0.01% of samples.
Our data show a substantial variation of approaches for DPYD genotyping used in laboratories across Europe as well as a limited use of second, independent techniques as a confirmation method to assure the correctness of genotyping results. Almost half of the responders do not apply any of the suggested confirmation or replication methods, and implies the need for centrally organized guidelines. We selected DPYD as an example for its clinical relevance, as a false negative result or misclassification can have a fatal outcome. The number of centres which routinely test for DPYD is relatively low and it is possible that a questionnaire focussing on a gene that is more commonly tested would have resulted in a higher response rate. However we do not expect major differences in confirmation practice between genes within a laboratory.
To assess the usefulness of applying two independent genotyping techniques for confirmatory testing we evaluated genotyping results of almost 90,000 samples tested in two laboratories in over ten years of genotyping. We identified nine discrepant results (0.01%) between the two independent genotyping techniques. One discrepant result was caused by chimerism following allogeneic hematopoietic stem cell transplantations, and is thus not due to analytical failure. To prevent this particular type of error, a check-box for “transplantation patient” was added to the genotyping request form. Two other stem cell transplantation patients were correctly genotyped after the check-box was added. For the other eight samples, misclassification due to allele dropout was the most probable cause of the discrepancies. In this study, a frequency of 0.01% of misclassification was shown, whereas previous publications show higher frequencies of misclassification (0.27% in 365 patients, Scantamburlo et al.29 and 0.44% in 30,769 genotypes, Blais et al.).24 A difference in discrepant results between the two genotyping centres was identified and might be explained by the different genotyping techniques used in each centre, as the call rate and accuracy of the techniques can be different. Additionally, CYP2D6 data of one centre was not included, as this centre did not use a second, independent genotyping platform to confirm genotyping results for CYP2D6. CYP2D6 is a highly polymorphic gene and CYP2D6-assays could be more prone to allele dropout.
Another important aspect to consider is that allele dropout is test specific: it depends on the positions of variants and the primer positions of the assay. Therefore, caution should be taken in generalizing our results. Specific quality control analyses per assay may be warranted. One could envision for example a minimum amount of samples to be tested to show that allele dropout for that particular assay and primers is low, possibly as a requirement for diagnostic companies to demonstrate. This brings along a second important consideration, which is that the sensitivity of detecting allele dropouts is directly
234