Page 258 - Personalised medicine of fluoropyrimidines using DPYD pharmacogenetics Carin Lunenburg

Chapter 10
supplementary material. In addition, the time of last food intake prior to the blood draw for the endogenous DHU/U ratio was collected in all patients.
Sample size calculation and statistical analyses
The sample size for comparison of the four phenotyping assays was based on the co-primary aim (the association between the result of a phenotyping assay and severe fluoropyrimidine- induced toxicity). It was calculated to be 240 patients (see detailed description in the supplementary material). The association between the result of a phenotyping assay and DPD deficiency (as determined by the DPD enzyme activity in PBMCs) was investigated as secondary aim.
Patient characteristics or toxicity differences between patient groups were tested using χ2 test or non-parametric test. The DPD phenotyping assays were executed for the first time in the participating centres, i.e. in a research setting and were not cross-validated. To investigate the effect of centres on the outcome per assay, a mixed model analysis was executed to demonstrate the general reliability of the assay. A univariate analysis of variance was done to compare outcomes per centre. In this analysis, the centre with the highest recruitment rate was chosen as the reference centre. DPYD variant allele carriers were excluded in these analyses, as they are underwent a per protocol dose adjustment and hence are at lower risk for toxicity. Age, gender and baseline BSA were additionally taken into account as possible covariates to minimize the risk of biased results of the analyses. When age, gender or baseline BSA were associated with the outcome of the phenotyping assay, the distribution of the covariate was checked between centres using Chi-Square tests or univariate analysis of variance. The possible clinical consequence of the divergent results was discussed per assay and data could be excluded from further analyses.
For assessing clinical validity, measures to determine diagnostic performance (i.e. sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV] and F1-score [harmonic mean of sensitivity and PPV]) of the assays with regard to the onset of severe toxicity or DPD deficiency were determined. DPYD variant allele carriers who received dose reductions based on their identified genotype could only be included in an analysis for association with DPD activity, not for the association with onset of severe toxicity. The level of significance was set at p<0.05. Analyses were performed using SPSS, version 23 (IBM SPSS Inc., Chicago, IL, USA).
Results
Patients
In total, 1,037 patients participated in the phenotyping study of which 1,037, 92 and 82 patients underwent two, three and four phenotyping assays, respectively. Patient and treatment characteristics of the 92 patients were similar to those of the main study (N=1,103), with the exception that the 92 patients were slightly younger (median age of 60 versus 64 years, p=0.011, Table 1). Details on fluoropyrimidine-induced toxicity are depicted in Supplementary Table 1. In total, 19 out of 92 patients (21%) experienced severe fluoropyrimidine-induced toxicity, which is comparable to the main study in which 264 out of 1,103 patients (24%) experienced severe toxicity (p=0.477).
256