Page 265 - Personalised medicine of fluoropyrimidines using DPYD pharmacogenetics Carin Lunenburg

Phenotyping assays for predicting DPD deficiency
The goal of this study was to explore the clinical value of DPD phenotyping assays to identify DPD deficient patients with an increased risk for severe fluoropyrimidine-induced toxicity. Previously, high endogenous uracil levels have been associated to the onset of severe toxicity.20 In our study, there was no difference between the mean endogenous DHU/U ratio or mean endogenous uracil levels between wild-type patients for four DPYD variants, who experienced severe toxicity or not. Possibly, when including DPYD variant allele carriers a difference would have been visible. Yet, we aimed to identify DPD deficient patients in addition to DPYD variant allele carriers who are DPD deficient. In terms of clinical validity parameters, our results for the endogenous uracil levels (sensitivity 21%, specificity 82%, NPV 77%, PPV 27%) were only slightly different from previously published parameters to predict severe fluoropyrimidine-induced toxicity (sensitivity 18%, specificity 95%, NPV 90%, PPV 35%).20 Taking the limited number of patients for two out of four phenotyping assays into account, none of the phenotyping assays investigated in this study showed a combination of both high PPV and NPV parameters, which could predict the clinical value of a test. Of note, sensitivity and PPV of an assay will remain limited even though there is a high odds ratio, if e.g. adverse events are frequent and deficient patients are rare.32 This is also the case for DPD deficiency and severe fluoropyrimidine-induced toxicity, and we identified
low clinical validity parameters.
Our study is the first study in which several phenotyping assays were compared head-
to-head in the same patients. For two out of four assays (endogenous DHU/U ratio and endogenous uracil levels), we recruited over 1,000 patients representative of routine clinical care patients. However, our study has some limitations. The 92 patients who underwent three or four phenotyping assays were a little younger compared to patients from the main study cohort, possibly due to the higher patient burden to participate in multiple DPD phenotyping assays. However, we feel this difference is not clinically relevant and it did not influence the occurrence of severe fluoropyrimidine-induced toxicity in these patients.
Secondly, we identified variation in the results of the phenotyping assays, either possibly caused by differences between centres or baseline characteristics (i.e. age, gender or BSA). Per assay, we have examined divergent results and we have corrected for these variations by excluding patients. While we now attributed the identified variation to differences between study centres, these divergent results might also be caused by already existing fluctuation in phenotyping results due to the character of the assay and measurement method. In addition, variation in the clock time of sampling may have affected uracil levels as the metabolizing enzyme DPD shows significant circadian variation.11 Variation in phenotyping results might also be caused by a different distribution of DPD deficient patients between centres.
Furthermore, predefined cut-off values per phenotyping assay derived from literature were used to be able to divide patients in DPD deficient and non-DPD deficient patients and calculate clinical validity parameters. Cut-off values are also necessary for clinical use, as it would be difficult to determine which patients would require an initial dose reduction without the use of cut-off values. However, the use of cut-off values limits the interpretation of the data of each phenotyping assay. In addition, DPD deficiency itself does not follow a cut-off at a certain point as its severity varies in gradation between completely DPD deficient, partially DPD deficient or non-DPD deficient. Therefore, it would be better not to use cut-off
10
263