Considering the use of biomarkers: a critical appraisal

Even with the use of a genotypic tests that consider multiple DPYD variants simultaneously, some argue that measures of clinical validity for these biomarkers are too low to justify routine use. The aim of using these biomarkers is to prevent severe and potentially lethal toxicity, but the sensitivity and positive predictive value of the genotype tests remain stubbornly low because not all toxicity can be explained by DPYD variants.The specificity and negative predictive value is more impressive, but is not 100%. The risk here is that some patients will have their dose modified and not get the full benefit of a standard chemotherapy dose (even though they would not experience toxicity). In practice, optimal doses can be achieved through the use of a cautious starting cycle, which is followed by dose escalation.

Phenotypic Testing: are phenotypic tests better than genetic biomarkers?

Phenotypic tests measure DPD activity more directly. The following phenotypic tests could predict DPD deficiency and therefore improve the validity profile of biomarkers:

Phenotyping Method	Comments on validity/utility/feasibility
DPD enzyme activity in peripheral blood mononuclear cells	⚬ The gold-standard for phenotyping. ⚬ Can identify patients heterozygous for DPYD variants. ⚬ Procedure is time-consuming.
Endogenous uracil (and dihydrouracil) concentrations	⚬ Moderate correlation with DPD activity. ⚬ Large overlap with control population.
Uracil Breath Test	⚬ Can identify patients heterozygous for DPYD variants. ⚬ Procedure is expensive.
Uracil Test Dose	⚬ Can identify patients heterozygous for DPYD variants. ⚬ Procedure is cheap and easy to implement.

Pharmacokinetic and Pharmacodynamic Pathways

Some patients without a variant DPYD allele and who are not outliers with respect to DPD activity may still experience severe toxicity or indeed fail to respond to 5-FU. This can be due to a range of other genetic (constitutional and somatic) or environmental factors.These images illustrate the pharmacokinetic and pharmacodynamic pathways for 5-FU and its prodrugs at the molecular level^{Image 1. Fluoropyrimidine Pathway, Pharmacokinetics (Click to enlarge) Image 2. Fluoropyrimidine Pathway, Pharmacodynamics (Click to enlarge)}Please note some of the factors illustrated here and other factors that could affect 5-FU response:

• Response to 5-FU and the prodrugs can also be influenced by variants in many well-studied genes, including TYMS which encodes the primary pharmacodynamic target.
• Somatic tumour mutations including those to drug targets, such as amplification of TYMS, will have an impact on treatment response.
• There are other rate-limiting steps in 5-FU metabolism, in addition to the step dependent on DPD activity.
• Fluoropyrimidines are used in combination with various other antineoplastic drugs; in regimes that are specific to the disease itself. The particular regime will also influence the relative importance of DPD/DPYD variants and the risk of toxicity.

This complexity has made the development of a predictive testing strategy with evidence of clinical utility a particular challenge.