DPD identification is key in avoiding serious reaction to 5-FU cancer drug

Before starting cancer treatment with fluoropyrimidine-based chemotherapies, it is highly recommended to check for dihydropyrimidine dehydrogenase (DPD) deficiency by measuring uracilemia (or calculating the dihydrouracil:uracil ratio). This article discusses some of the ways of doing this.
Background
Approved for treatment of humans 60 years ago, fluoropyrimidinebased chemotherapies remain important antineoplastic agents. They are widely used in Europe, for example in France 100¦000 patients are medicated with this group of anticancer drugs.
Indeed, 5-fluorouracil (5-FU) and its oral pre-prodrug capecitabine are the backbone in the treatment of colorectal, pancreatic, gastric, breast, head and neck cancers. They work by interfering with enzymes (principally thymidylate synthase) involved in producing new DNA, thereby blocking the growth of cancer cells. They are administered by injection or by mouth. However, the use of fluoropyrimidines is associated with an important risk of toxicity, mainly due to deficiency of the enzyme involved in its catabolism, dihydropyrimidine dehydrogenase (DPD).
In France, health authorities recommend the determination of uracil concentration to guide dosing of fluoropyrimidines. Numerous liquid chromatography-tandem mass spectrometry (LC-MS/MS) methods have been proposed but they include complex liquid–liquid or solid-phase extraction procedures.
Prescribers may be unaware that their patients lack functional DPD (encoded by the DPYD gene) and hence cannot break down fluorouracil, resulting in its build-up. This can lead to severe and life-threatening side effects such as neutropenia, neurotoxicity, severe diarrhea and stomatitis.
Up to 15% of patients exhibit a partial deficiency, whereas 0.1–0.5% may have a complete deficiency. Consequently, a 5-FU dose can lead to severe or lethal toxicity, and it is therefore highly recommended to screen for DPD status to determine a safe dose for the patient.
This deficiency may be detected either by genotyping (an approach that explores the polymorphisms of the DPYD gene) or by phenotyping, which consists of measuring uracilemia or calculating the 5,6-dihydrouracil:uracil (UH2:U) ratio.
Brief methodological overview

• The genotyping approach explores four variants known for reducing DPD activity (DPYD*13, DPYD*9A, DPYD*2A, and 2846A>T) and has the advantage of producing a fast and relatively inexpensive response by using automated techniques. Its specificity is very good, but its sensitivity is poor (not all DPD deficiencies are detected by genotyping).
• DPD is essential for converting endogenous U to UH2. Therefore, uracilemia or the UH2:U ratio reflect the level of DPD activity. Measurement of these components is feasible in plasma by liquid chromatography with photodiode array detection (LC-DAD) and LC-MS but requires complex sample preparation with protein precipitation, liquid–liquid extraction (LLE) or solid-phase extraction. Up to now, only analytical methods with multiple manual steps involving centrifugation, filtration and evaporation have been reported. Although results are satisfactory, the methods are time-consuming and tedious.

In genotyping, genes causing the deficiency are focused on, whereas with LC-MS/MS, the activity of DPD is estimated by measuring the ratio of the compounds UH2 and U. The first method looks only at the cause, whereas the second, safer method, looks at the result considering all deficiency cases while reducing toxic risks.
Need for accuracy, reliability and robustness
Proposed threshold values of 16 and 150 ng/mL for uracilemia characterize a partial or complete DPD deficiency, respectively. Inaccurate quantification of these threshold values may totally influence patient care and medical decisions. Analytical methods must therefore be accurate, reliable and robust. Automation is undoubtedly the best solution for reduction of errors while ensuring best reproducibility, robustness and reliability.
In this context, Shimadzu has developed a fully-automated procedure for the measurement of U and UH2 in human plasma. It is known as indirect phenotyping and provides faster testing as well as greater accuracy, safety and standardization. It is a method where the extraction is carried out by a programmable liquid handler directly coupled to a LC-MS/MS system.
The Centre Hospitalier Universitaire de Limoges (CHU Limoges), France, has been involved in proposing a method combining accuracy and time-efficiency. They suggested a new solution based on a novel sample preparation system, coupling an HPLC instrument and a triplequadrupole mass spectrometer.
Extraction is performed by an automated sample preparation system, the Clinical Laboratory Automation Module (CLAM)-2030 (Shimadzu Corporation) coupled to an LC-MS/MS system. Responding to the needs of clinical research sites, the CLAM-2030 provides stable data acquisition, lower running costs and improved work efficiency. It can be connected to four models of triple-quadrupole liquid chromatography mass spectrometers. Once the primary (or secondary) tube is loaded onto the automated system, no further human intervention is required as the CLAM-2030 resulting in high standardization.
The system was used in positive electrospray ionization mode. Acquisition method targeted multiple reaction monitoring (MRM) transitions for uracil, dihydrouracil, uracil-13C, 15N2 and dihydrouracil-13C, 15N2. The workflow procedure is summarized in Figure 1.
The CLAM-2030 targets pharmaceutical and medical departments as well as biological analysis labs. It is a technological key system applied in Shimadzu’s European Innovation Center (EuIC) programme. The EuIC merges the cutting-edge analytical technologies of Shimadzu with game-changing topics and expertise in markets and science covered by opinion leaders, strategic thinkers and scientific experts in order to create new solutions for tomorrow. In France, the CHU University Hospital is a cooperation partner of the EuIC.
The CLAM-2030 module automates everything from the preparation of urine, blood, and other biological samples to measurement via liquid chromatography mass spectrometry (LC-MS). Within a few minutes, the CLAM-2030 preparation module completes the blood-sample preparation process including the addition of reagents, mixing of the solution and the addition of a deproteinization liquid, compared to the 15–20 minutes that this process conventionally takes. Further, if the samples and reagents are placed and positioned in special containers for automatic conveyance to the LC-MS by an autosampler, the module can perform all of the processes automatically, on weekends and overnight.
Quick results
By overlapping sample treatment, a result is obtained every 14 minutes after the first sample. This method is fully validated according to ISO 15189 requirements. The result of the validation study are summarized in Table 1. A 5 ng/mL limit of quantification is obtained for both U and UH2 with good linearity (R² >0.995). At 16 ng/mL (threshold value) the inaccuracy and coefficients of variation were less than 5% for intra- and inter-assay tests, clearly sufficient to avoid misdiagnosing the level of DPD activity.
The method has been applied successfully in 64 consecutive patients tested at the CHU Limoges, and its results were similar to those of a classic LC-MS method (LLE for sample preparation) used routinely until then. For each patient, the same diagnosis (absence or presence of DPD deficiency) was given and the Bland–Altman plot (Fig. 2) shows good agreement between the two methods.
Conclusion
As DPD deficiency screening in patients given fluoropyrimidine-based chemotherapy is now highly recommended, most labs in charge of the measurement of U (and UH2) will or are already facing an increase in this activity. Shimadzu therefore proposes a fully-automated solution ensuring an accurate and robust measurement without requiring precious laboratory staff time. The simplicity of operation and the minimization of user involvement in the sample preparation process will help obtain high throughput for the monitoring of 5-FU and capecitabine treatments.