Page 29 - Personalised medicine of fluoropyrimidines using DPYD pharmacogenetics Carin Lunenburg

Prospective DPYD genotyping: ready for prime time
In a previous study approximately 10% of the DPYD*2A variant allele carriers treated with
the standard fluoropyrimidine dose deceased as a result of drug-induced severe toxicity.35
The approach of pre-treatment genotyping followed by a reduced starting dose plus 2 tolerance-guided dose titration could prevent the occurrence of severe toxicities in DPYD variant allele carriers, resulting in a direct safer use with minimum risk of underdosing. The above mentioned test characteristics are reached using the two most investigated SNPs and
these values will probably improve when a larger panel of DPYD SNPs is probed. Costs are
not likely to increase substantially when adding SNPs because genotyping costs continue to decrease.36,37 Although more DPYD variants that alter DPD enzyme activity are continuously discovered and studied, the perfect set of SNPs has not been defined yet. Currently we
feel there is substantial evidence to support dose recommendations for at least four variants (DPYD*2A, c.2846A>T, DPYD*13 and c.1236G>A/HapB3).38 Another possibility for prospective screening could be the more informative, but hugely more expensive genotyping
of the entire coding region of DPYD. However we have focused on genotyping SNPs. To
date, SNP genotyping has been most extensively studied, is technically feasible in a general hospital setting and multiple guidelines providing SNP-based dose recommendations are available.
What is needed for implementation of DPYD genotyping in daily routine clinical care? Clinical implementation of a biomarker test such as DPYD pharmacogenomics is hampered due to the on-going discussion on whether a randomised clinical trial (RCT) is considered necessary to provide the required evidence before clinical implementation.26,29,37,39-45 Despite the fact that RCTs are considered the gold standard study design to prove effectiveness, adequate evidence can also be provided by small-scale, innovative, prospective interventional studies.40 However, with the available evidence favouring upfront genotyping, it may not be ethically feasible to randomise patients, and patients may not be willing to be included in the control arm with an increased risk for severe toxicity. Indeed, the only attempt at a prospective randomised study was performed in France. Boisdron-Celle et al. presented a multicentre prospective cohort study of upfront DPD deficiency screening executed from 2008 until 2012.46 The purpose of the study was to confirm the medical and economic aspect of upfront DPD deficiency screening in a prospective way as was done retrospectively by Traoré et al.47 Patients using 5-FU based chemotherapy were included in one of two parallel patient cohorts (arm A and arm B). Patients in arm A were prospectively screened for DPD-deficiency (a combined genotyping and phenotyping approach), and patients in arm B were retrospectively tested. A total of 1,130 patients were included (arm A: 720 patients, arm B: 410 patients). One patient died due to 5-FU early-onset toxicity and it was retrospectively confirmed that this patient was DPD deficient (arm B). The enrolment of patients was prematurely closed for ethical reasons, because of the proven 5-FU-induced toxic death of this patient.46,48 Against this background, we conclude that evidence from a randomised prospective clinical trial on DPYD genotyping will never be acquired for ethical reasons. In addition, some predictive biomarkers were previously implemented without evidence from an RCT. Clinical use of (K)RAS selection for EGFR therapy was influenced by updated registration texts for epidermal growth factor receptor (EGFR) inhibitors from
27