Page 28 - Personalised medicine of fluoropyrimidines using DPYD pharmacogenetics Carin Lunenburg

Chapter 2
for the use of DPYD genotyping prior to start of treatment with fluoropyrimidines.27,28 Other DPD deficiency screening methods (e.g. phenotyping) have been described,29 and are currently being investigated (NCT02324452), but we feel are not ready yet for clinical application. In the current paper, we present an overview on the evidence for prospective DPYD genotyping and discuss critical questions related to its implementation. Associations of DPYD variants with fluoropyrimidine-induced toxicity, prevention of severe toxicity upon DPYD testing, cost consequences and existing guidelines will be discussed.
Available evidence for the association of DPYD variants and 5-FU-induced severe toxicity The relationship between DPYD variants and 5-FU-induced severe toxicity is widely acknowledged. Recently, data have been summarised in three separate meta-analyses.8,9,30 Terrazzino et al. evaluated 4,094 patients (15 studies) for DPYD*2A (IVS14+1G>A; rs3918290) and 2,308 patients for c.2846A>T (D949V, rs67376798). They confirmed the clinical validity of these SNPs as risk factors for the development of fluoropyrimidine-associated severe toxicities (details in Table 1).9 The second meta-analysis, performed by Rosmarin et al., included data of 4,855 patients (17 studies). They describe eight DPYD variants of which DPYD*2A and c.2846A>T also showed convincing evidence of an association with toxicity (Table 1).8 The third meta-analysis of Meulendijks et al., included data of 7,365 patients (eight studies) and confirmed the association between severe toxicity and the variants DPYD*2A and c.2846A>T, but also for DPYD*13 (I560S; c.1679T>G; rs55886062) and c.1236G>A/ HapB3 (E412E; rs56038477) (Table 1). Very recently, three additional papers, not part of the three meta-analyses, have confirmed significant associations between DPYD variants and toxicity (Table 1).4,31,32 Although multiple variants of DPYD have been described, DPYD*2A, DPYD*13, c.2846A>T and c.1236G>A/HapB3 are the variants that are most extensively studied and convincingly associated with fluoropyrimidine-related severe toxicity.8,9,30
The HuGE risk translator33 is an online tool to calculate test characteristics for the evaluation of the predictive ability of genetic markers. Data (e.g. odds ratio) from two of three meta-analyses described above could be entered as a ‘two-risk genotype’ for DPYD*2A and c.2846A>T, resulted in low (~10 to ~25%) sensitivity and positive predictive values and high (>96%) specificity and negative predictive values (NPV). The number needed to screen (i.e. genotype) appears to be 210–250 patients and the number needed to treat (i.e. apply dose adjustments) is five or six patients (Table 2). Important to note is that values for diagnostic test criteria of a pharmacogenomic test based on SNPs in DPYD can never reach 100%, because not all DPD deficiencies and toxicity can be explained by variants in DPYD.34 It must also be said that the high specificity (±98%) and high NPV (±96.5%) in this setting are most important, when the goal is to treat all patients with a variant (including false-positives). The consequence of a (false) positive result is a relatively low-risk dose- reduction for the first of many cycles, which can be adjusted in safe conditions in the second cycle and onwards if no toxicity occurs. The consequence of a false negative result may be much larger since it could result in a too high systemic drug exposure that subsequently leads to severe, potentially lethal toxicity, which is associated with long-lasting hospital and/ or intensive care unit (ICU) admissions.
26