It is estimated that two million cancer patients worldwide are treated with fluoropyrimidines every year. This group of drugs includes 5-fluorouracil (5-FU) and capecitabine, the latter being a prodrug metabolized into 5-FU (Longley et al., 2003; Walko and Lindley, 2005). Fluoropyrimidines are a cornerstone of chemotherapy, commonly prescribed for adjuvant and palliative treatment of multiple solid tumour types, including colorectal, head and neck, gastric, pancreatic, and breast cancers. However, 10%–40% of the treated population develop severe treatment-related toxicity, leading to hospitalization, treatment discontinuation, or even death in approximately 1% of patients (Hoff et al., 2001; Twelves et al., 2005; André et al., 2015; Amstutz et al., 2018). 10% of patients develop these severe toxicities within the first three chemotherapy cycles and are obliged to discontinue therapy (Meulendijks et al., 2016b). The most common fluoropyrimidine-induced adverse events are diarrhea, nausea, vomiting, mucositis, myelosuppression, and hand-foot syndrome (HFS) (Negarandeh et al., 2020).

One of the best-studied and recognized causes of fluoropyrimidine-related toxicity is the reduced activity of the primary 5-FU inactivating enzyme dihydropyrimidine dehydrogenase (DPD), caused by genetic variants in DPYD, the gene encoding DPD. The interindividual variability in fluoropyrimidine metabolism by DPD results in toxic concentrations of drug exposure in some patients and not others. It has been repeatedly demonstrated in retrospective and prospective studies that patients with DPD deficiency receiving standard doses of fluoropyrimidine therapy are at a significantly higher risk of severe toxicity (van Kuilenburg, 2004; Terrazzino et al., 2013; Rosmarin et al., 2014; Meulendijks et al., 2015; Toffoli et al., 2015; Barin-Le Guellec et al., 2020; Deng et al., 2020). The most well-studied polymorphism is DPYD*2A (c.1905 + 1G>A), which results in a complete loss of function. Approximately 1.6% of individuals of European ancestry are heterozygous carriers of DPYD*2A and thus have a 50% reduction in DPD function relative to those with two completely functional alleles (Deenen et al., 2016; Amstutz et al., 2018). Additionally, there are three other DPYD variants with a statistically significant association with severe 5-FU toxicity that are deemed clinically relevant: the loss of function allele DPYD*13 (c.1679T>G) as well as the reduced function alleles c.2846A>T and c.1236G>A (HapB3) (Amstutz et al., 2018; Henricks et al., 2018). Together, these four alleles may be found in 3%–8% of those of European ancestry. Expert groups have recommended testing for these four variants (Amstutz et al., 2018; Hamzic et al., 2020a; Lunenburg et al., 2020).

Many other DPYD variants that impact DPD phenotype in vitro are extremely rare and have not been identified in large cohort studies (Offer et al., 2014; Amstutz et al., 2018). Two additional variants were considered due to the ancestrally diverse Vancouver population. Fluoropyrimidine toxicity may be seen in patients carrying c.557A>G, a reduced function variant present in 3% of individuals with African ancestry, as well as c.2279C>T, a reduced function variant present in 1% of individuals with South Asian ancestry (Offer et al., 2013; 2014; Elraiyah et al., 2017).

In a prospective trial that screened for DPYD*2A in 2038 patients, Deenen et al. demonstrated that a 50% dose reduction for variant carriers reduced the frequency of severe fluoropyrimidine-related toxicity from 73% to 28%, which was comparable to the 23% rate of toxicity observed in non-carrier patients treated with a full dose (Deenen et al., 2016). In another large prospective trial, Henricks et al. screened for all four aforementioned “European” DPYD variants. They found that an upfront dose reduction of 50% in heterozygous c.1905 + 1G>A carriers reduced the frequency of severe toxicity from 77% to 18% (Henricks et al., 2018). Importantly, Henricks et al. also found that a 25% dose reduction in c.1236G>A and c.2846A>T carriers was insufficient, as the risk of severe toxicity (39% in c.1236G>A and 47% in c.2846A>T) remained elevated relative to non-carriers of the DPYD variants. Consequently, the fluoropyrimidine dosing guidelines by the Clinical Pharmacogenetics Implementation Consortium (CPIC) now recommends a 50% initial dose reduction for patients who are heterozygous for either reduced function or non-functional DPYD variants (Amstutz et al., 2018).

Importantly, these trials demonstrated that dose reductions of capecitabine and 5-FU in DPYD variant carriers likely did not underdose patients. This has been demonstrated by pharmacokinetic analyses showing similar levels of drug exposure in heterozygous carriers treated with reduced doses and non-carriers treated with standard doses (Henricks et al., 2018). Moreover, retrospective studies have shown that DPYD variant carriers experience up to an 88% increased risk of grade ≥ 3 fluoropyrimidine toxicity (van Kuilenburg et al., 2000; Van Kuilenburg et al., 2002). Given a relatively high carrier frequency, a large group of at-risk patients can be pre-emptively identified through prospective genotyping (Amstutz et al., 2018). Prospective data show lower toxicity with properly informed genotype-guided dosing with no consequent decrease in survival (Henricks et al., 2019). Furthermore, it has been shown that genotype-guided dosing results in equivalent systemic fluorouracil concentrations across all dosing schedules (Deenen et al., 2016; Henricks et al., 2018; 2019).

Currently, routine prospective DPYD genotyping is not the standard of care in North America, although recent implementation trials have emerged (Reizine et al., 2020; Varughese et al., 2022; Baker et al., 2023). The National Comprehensive Cancer Network (NCCN), the American Society of Clinical Oncology (ASCO), and the US Food and Drug Administration (FDA) do not explicitly support routine pre-treatment DPYD testing. However, it is important to also note that not every agency agrees that the lack of pre-emptive testing guidance from NCCN or the labeling by the FDA is appropriate. The European Medicines Agency, the French National Agency for the Safety of Medicines and Health Products, and the Medicines and Healthcare Products Regulatory Agency (United Kingdom) have each approved guidelines for pre-emptive determination of DPD activity for patients treated with fluoropyrimidines (Medicamentet des Produits de Sante, 2019; European Medicines Agency, 2020; NHS England, 2020; UK Chemotherapy Board, 2020). Expert commentaries recently published discuss the lack of support for DPYD testing before fluoropyrimidine chemotherapy in North America (Hertz, 2022; Baker et al., 2023; Hertz et al., 2023). They cite the FDA labelling which acknowledges that patients with DPD deficiency have an increased risk of life-threatening toxicity, but instead of recommending testing, FDA suggests an unlikely scenario in which patients who have known DPD deficiency should discuss it with their physicians. The authors are oncologists, clinical pharmacists, and other experts affiliated with various oncology centres and conclude with a recommendation for standard of care pre-treatment DPYD testing and urging NCCN, ASCO, and FDA to update their guidelines. It is also notable that the Dana Farber Cancer Institute (Boston, United States) launched a program to promote DPYD testing in 2022; Oregon Health and Science University paid USD$1M to settle a lawsuit from a widow of a patient after DPD deficiency was implicated in the fatality.

In Canada, other methods to determine DPD function such as therapeutic drug monitoring (TDM) and phenotyping have not been implemented to our knowledge. In the absence of prospective screening, physicians are blind to the necessity of dose reduction, inevitably resulting in an avoidable increase in the incidence of severe toxicity. Prior to this initiative, prospective DPYD genotyping was not routinely available in British Columbia (BC). As such, patients predisposed to fluoropyrimidine toxicity could not be identified prior to initiation of therapy, resulting in a preventable increase in morbidity and mortality.

A testing program was developed for genotype-guided dosing of fluoropyrimidines based on six variants (Table 1). This program was successfully implemented at BC Cancer Vancouver Centre in August 2022 and served as a model for the program’s provincial expansion in May 2023. The program’s central aim was to prevent severe fluoropyrimidine-related adverse drug reactions through DPYD genotype-guided dose individualization. An additional aim was to provide prospective data upon which to refine genotype-based dosing strategies, including collecting prospective data for DPYD variants prevalent in those of non-European ancestries (c.557A>G, c.2279C>T).

Between September 2022 to May 2023, 186 patients received DPYD testing (Table 3). The most common tumour types were colorectal (n = 80; 52%), breast (n = 21, 14%), and pancreatic (n = 18, 12%). The most common self-reported ancestries were European (n = 84; 55%) and East Asian (n = 26, 17%). There were 14 (8%) heterozygous DPYD variant allele carriers. The cohort had no significant differences in sex, age, tumour type, tumour stage, and treatment regimen between DPYD variant and non-variant carriers.

Out of the 14 DPYD variants identified in this cohort, four were retrospectively discovered after discontinuing chemotherapy due to fluoropyrimidine toxicity (Figure 2). Out of the remaining 10 prospectively identified variant carriers, three DPYD variant carriers received their results after initiating chemotherapy. One patient tolerated the first cycle at the full dose, but the physician opted to discontinue fluoropyrimidine-based therapy, citing limited benefit and increased toxicity risk with future cycles. The other two patients received guideline-recommended dose reductions in cycle 2 after their results were returned and one experienced severe toxicity (CTCAE grade 3) despite the reduced dose. As these two patients promptly received dose reductions after a cycle, they were included in the analysis as dose-reduced DPYD variant carriers (Table 4). Seven DPYD variant carriers received their pharmacogenetic reports before the first cycle of chemotherapy and were dose-reduced by 50% for cycle 1.

In the first three cycles, toxicity outcomes were compared between non-variant carriers and variant carriers who had treatment adjusted based on genotype (Table 4). All nine dose-reduced variant carriers and 127 (74%) of 178 non-variant carriers had completed three cycles of chemotherapy or completed their regimen before cycle 3 (Figure 2). Variant carriers given DPYD results received a median relative dose intensity of 50% (IQR 45%–74.5%). Most non-variant carriers received the standard dose, with a median relative dose intensity of 100 (IQR 92%–100%).

Of nine DPYD variant carriers receiving DPYD-guided fluoropyrimidine dose reductions, two experienced severe fluoropyrimidine-related toxicity compared to 14% in non-variant carriers receiving standard doses (Table 4). One (11%) variant carrier receiving a 50% dose reduction was hospitalized due to fluoropyrimidine-related toxicity, while six (5%) non-variant carriers were hospitalized.

DPYD variant carriers are more vulnerable to adverse fluoropyrimidine reactions due to being exposed to elevated systemic levels of 5-fluorouracil (Amstutz et al., 2011). Although no pharmacokinetic analyses were done for this study, other pharmacokinetic studies from prospective DPYD testing trials have shown that dose reductions help establish therapeutic concentration levels, allowing variant carriers to be treated with fluoropyrimidines at an exposure comparable to non-variant carriers receiving standard doses (Deenen et al., 2016; Henricks et al., 2018). However, the authors noted that there is large interindividual variability in the exposure of capecitabine and its metabolites, so pharmacokinetic results should be interpreted cautiously (Meulendijks et al., 2015).

DPYD-guided dose reductions allowed seven of nine variant carriers to be more safely treated with fluoropyrimidines and therefore likely prevented dose-related toxicity. The incidence of severe toxicity in DPYD variant carriers treated at standard doses is over 70% (Meulendijks et al., 2015; Deenen et al., 2016). In this study, two out of nine DPYD variant carriers still experienced grade 3 toxicity. A DPYD-guided dose reduction was especially significant for the patient hospitalized after just 8 days of initiating capecitabine. Even at a 50% dose, the patient’s platelet count dropped 82% to 35 × 10⁹/L from a normal baseline. At a greater than 50% dose reduction, this patient was able to finish their course of adjuvant capecitabine with concurrent radiotherapy. For anal squamous cell cancer, pre-operative capecitabine and radiotherapy treatment were important to achieve a complete clinical response, which studies have shown to happen in 89.7% of patients (Meulendijks et al., 2014). This case influenced oncologists’ views on the necessity of DPYD testing and dose reductions for variant carriers. Most did not initially believe that pre-treatment DPYD testing was necessary, a sentiment echoed in a survey of US medical oncologists where only 32% of responders would recommend testing (Koo et al., 2022). However, due to the severity of this patient’s toxicity within the first week of treatment, oncologists accept that DPYD testing likely prevented lethal fluoropyrimidine toxicity. As a group, oncologists are now more willing to follow dose reduction guidelines for variant carriers. Critical to the use of guidelines were reminders that fluoropyrimidine patients who are variant carriers likely have higher toxic concentrations than those who are not variant carriers.

In non-variant carriers, 14% experienced severe fluoropyrimidine toxicity, which was in the expected range of 10%–30% (Mikhail et al., 2010; Froehlich et al., 2014; Meulendijks et al., 2016b; Barin-Le Guellec et al., 2020). The most common type of toxicity experienced was gastrointestinal toxicity (8%), then hematological toxicity (5%) (Table 4), frequencies also reflected in the literature (Meulendijks et al., 2016b). DPYD variants are known to be a major contributor to severe fluoropyrimidine toxicity in the first three cycles (Froehlich et al., 2014). DPYD variant carriers were not only more likely to experience adverse fluoropyrimidine events, but they also experienced toxicity in multiple categories, compounding the risk of therapy delay and cessation. This, taken into account, along with the fact that two out of nine variant carriers experienced grade 3 toxicity in cycles 1 and 2 despite reduced doses, reflects how variant carriers are substantially more at risk for fluoropyrimidine toxicity (Froehlich et al., 2014).

After DPYD-guided dosing, guidelines suggest subsequent titrating subsequent doses to toxicity (Amstutz et al., 2018). However, oncologists requested further guidance on dose escalations in variant carriers who tolerated reduced doses. One of the patients experienced nausea in cycle 3 after a dose escalation from 50% to 67%, which was resolved through anti-emetics. Another patient, whose DPYD status was unknown at the time of treatment, experienced severe diarrhea after their capecitabine dose was increased from 50% to 75% dose. If this patient’s variant carrier status had been known beforehand, a smaller dose increase might have been more successful. One study examined tolerance-based dose escalations of capecitabine, where 11 heterozygous DPYD variant carriers tolerated a mean dose escalation of 8.5% (Kleinjan et al., 2019). More experience and data need to be collected on tolerance-based fluoropyrimidine dose increases in DPYD variant carriers to inform physicians on how much they can escalate the dose. Given the severity of toxicities and the risk of not being able to resume chemotherapy treatment, the guidelines were modified to suggest a dose increase of 10% if a variant carrier tolerated a reduced dose without toxicity for two cycles.

The above notwithstanding, the authors acknowledge a lack of data on dose escalations for DPYD variant carriers in the literature. Prospective DPYD-guided dose reduction studies did not report on the size of dose increases (Deenen et al., 2016; Henricks et al., 2018). Henricks et al. described that 5 out of 11 dose escalations for DPYD variant carriers were not tolerated, leading to subsequent dose reductions and treatment discontinuation. Although the sizes of dose increases were not mentioned, this data emphasizes the need to be cautious with dose escalations. There are studies that implemented a dose escalation algorithm through TDM with 5-FU. Kaldate et al. specified a maximum 30% dose increase per cycle if AUC is less than 8, with a target AUC of 20–30 (Kaldate et al., 2012). TDM algorithms have also not been established for capecitabine, and systemic levels of capecitabine and its metabolites were shown to be poor in predicting toxicity in phase III studies, leading the authors to recommend against TDM for capecitabine (Gieschke et al., 2003).

A limitation of this study was the lack of assessment of other personalized medicine approaches to determine DPD activity such as TDM and phenotyping. Used in tandem with genotyping, phenotyping and TDM could capture more patients at risk of fluoropyrimidine toxicity. TDM has been shown to be useful in improving the pharmacokinetic interindividual variability and safety of 5-FU (Yang et al., 2020). TDM could also prevent underdosing in dose-reduced DPYD variant carriers and would be a way to escalate doses in those that tolerate reduced doses safely. Challenges to widespread TDM implementation include the instability of 5-FU in blood and the variability in measurements due to differences in the timing of blood samples (Hamzic et al., 2018; Hamzic et al., 2020a; Schneider et al., 2021).

There are also challenges with phenotyping. In a prospective study of DPD phenotyping with plasma uracil levels, pre-treatment levels varied between institutions and did not correlate with the DPD activity in peripheral blood mononuclear cells (de With et al., 2022). Phenotyping involving peripheral blood mononuclear cells is also costly, time-consuming, and difficult to upscale due to its complexity and specialized equipment (Chazal et al., 1996; Knikman et al., 2021). Comparatively, DPYD genotyping is a simple and cost-effective first step to pre-emptively identify at-risk patients (Meulendijks et al., 2016a; Meulendijks et al., 2016b; Hodroj et al., 2021; Knikman et al., 2021; Schneider et al., 2021). Access to DPD phenotyping and TDM would have utility in refining fluoropyrimidine dosing, especially for rare patients with a DPD activity score of zero. None were observed in this study, but for these patients, clinical guidelines state to avoid fluoropyrimidines entirely. However, in cases when this is not possible, phenotyping and TDM may be helpful in determining a patient’s residual DPD activity (or ability to otherwise clear 5-FU) in order to establish a starting dose (Lunenburg et al., 2020).

In the prospectively tested cohort, 10 (5%) of patients were found to be variant carriers, a frequency that was expected based on the prevalence of variants in the testing panel (Amstutz et al., 2018). Previous studies reported the limitation that the guideline recommended DPYD variants to test for were predominantly validated in European populations. Therefore, in this study’s variant selection, variants impacting DPD function observed in non-European populations were added. This was the first DPYD study to prospectively test for c.2279C>T, a variant found in 1% of those with South Asian ancestry (Offer et al., 2014). This study was also the second to prospectively evaluate c.557A>G, a variant found in 3% of those with African ancestry (Offer et al., 2013; Elraiyah et al., 2017). This variant was included in another DPYD implementation protocol in the US (Reizine et al., 2020) but prospective results from DPYD testing trials have not been reported on this variant yet (Meulendijks et al., 2015; Boisdron-Celle et al., 2017; Wigle et al., 2021). One patient tested positive for c.557A>G, affirming the clinical value of adding it to the testing panel.

In this study, the testing program development and implementation took place at the same time, providing data that informed improvements in the workflow. Initially, physicians were not checking for DPYD status determination before initiating fluoropyrimidine-based chemotherapy. This was primarily due to adding a new testing program to an existing workflow that did not include testing. Oncologists were accustomed to booking bloodwork and chemotherapy for some patients before their first appointment. While a turnaround time of two days would eliminate treatment delays, this is not the most efficient and cost-effective way to offer testing to a large population. To work around this limitation, physicians, more familiar now with the testing process, have begun ordering tests before a patient’s first appointment and in a few cases, adjusting chemotherapy start dates by a few days to accommodate the turnaround time.

A single physician may treat many patients without a single positive test result, which could lead to reduced confidence in the benefits of genotyping (Begre et al., 2022). However, people do have rare variants and, as a result, suffer from severe toxicities. In this cohort, there was a DPYD*13 variant carrier, a variant with a prevalence of 0.01% in those of European ancestry (Table 1).

Proactive engagement between the research team and oncologists led to high enrolment in the study, which prospectively screened all BC Cancer Vancouver site patients who were expected to start on a fluoropyrimidine-based regimen. Having a research team member on site was crucial for identifying patients suitable for the study and encouraging enrolment. Due to the high volume of clinic patients starting fluoropyrimidines, the research team member was needed to actively screen and consent patients. An on-site research assistant was crucial in the first 9 months to ensure test uptake and standardized outcome reporting. When introducing a new test, success is enhanced by the presence of dedicated personnel to help integrate it into existing workflows until it becomes routine for clinicians. However, this role is likely not necessary for the long-term maintenance of a pharmacogenetic testing program. After 9 months, we transitioned to integrating test requisition links into the electronic medical record to remind physicians to order DPYD testing. Testing continues at the same rate and penetrance as before.

A strength of this study is that it is a real-world study that provides experience in pharmacogenetic testing implementation. However, oncologists were not blinded to patient genotype when collecting toxicity outcomes, potentially leading to bias in toxicity staging. In addition, as genotyping for only six DPYD variants was implemented, not all DPD-deficient patients could be captured. Now that an in-house genotyping platform has been established, there is an opportunity to add, in the future, other variants that may impact fluoropyrimidine toxicity. Of interest would be the inclusion of a genetic variant in TYMS (rs45445694), which encodes 5-FU’s target thymidylate synthase (TS). This variant has been associated with increased risk for the common capecitabine-related toxicity hand-foot syndrome (OR 1.32, p < 0.0001) (Hamzic et al., 2020b). Other variants could also explain the severe toxicity seen in the variant carrier in this study who already received the recommended dose reduction. However, the effect sizes are small and expert guidelines do not recommend them currently. It may be that in time, stronger evidence is forthcoming to justify inclusion. In addition, pharmacogenetic variants relevant to other medications patients often take during chemotherapy may be of interest, such as TPMT and NUDT15 testing to prevent thiopurine-induced myelosuppression, or CYP2D6 testing to prevent metoclopramide-related adverse events in poor metabolizers (Relling et al., 2011; Livezey et al., 2014).

A potential limitation of this study was the lack of a control group, making it more difficult to observe the benefits of DPYD testing. However, it was considered unethical to have a control cohort, as a previous prospective clinical study that had a group without dose individualization was terminated due to a DPYD variant carrier in the group dying from treatment toxicity (Boisdron-Celle et al., 2017). Although this study’s sample size did not allow for a statistically relevant number of dose-reduced variant carriers to compare to patients receiving the full dose, other large prospective studies have shown that DPYD-guided dose reductions in variant carriers result in similar toxicity levels to non-variant carriers receiving standard doses (Henricks et al., 2018; Wigle et al., 2021). Taken together, these results have led to the widespread adoption of DPYD testing by oncologists province-wide.

This study shows how DPYD testing can be implemented, integrated into an existing clinical workflow, and can become the standard of care. In conclusion, taken together with the primary literature, the results support that DPYD-guided dosing is a feasible approach to reducing the risk of severe fluoropyrimidine toxicity in DPYD variant carriers.