Drug monitoring detects under- and overdosing in patients receiving 5-fluorouracil-containing chemotherapy—results of a prospective, multicenter German observational study

Abstract

Introduction

Body surface area (BSA)-based dosing of 5-fluorouracil (5-FU) results in marked inter-individual variability in drug levels, whereas determination of plasma 5-FU concentration and area under the curve (AUC) is a more precise dosing method but has not been integrated into clinical routine. We conducted a multicenter, prospective study to study 5-FU AUC distributions and assess clinical factors predicting therapeutic dosing in patients receiving BSA-dosed 5-FU.

Methods

Between June 2017 and January 2018, a total of 434 patients receiving continuous, infusional BSA-dosed 5-FU from 37 sites in Germany were included. Plasma 5-FU concentration and AUC were measured in venous blood samples at steady state. The primary objective was to determine 5-FU AUC distributions in relation to the target range, which is defined as 20-30 mg × h/l. The second objective was to explore clinical parameters that correlate with achievement of 5-FU AUC target range.

Results

The primary tumor was mainly located in the gastrointestinal tract (96.3%), with colorectal cancer being the most common (71.2%) tumor entity. 5-FU was administered as monotherapy (8.1%) or as part of FOLFOX (33.2%), FOLFIRI (26.3%), or other regimens (12.4%). Treatment setting was adjuvant (31.3%) or metastatic (64.5%). The median AUC was 16 mg × h/l. Only 20.3% of patients received 5-FU treatment within the target range, whereas the majority of patients (60.6%) were underdosed and 19.1% of patients were overdosed. In the univariate logistic regression, treatment setting was the only clinical parameter that significantly correlated with achievement of the target range. Patients treated in the metastatic setting had a 2.1 (95% confidence interval 1.186-3.776, P = 0.011) higher odds to reach the target range compared with patients treated in the adjuvant setting.

Conclusions

The majority of patients received suboptimal doses of 5-FU using BSA dosing. Therapeutic drug monitoring of 5-FU is an option for optimized individualized cancer therapy and should be integrated into the clinical practice.

Highlights

• •BSA-based dosing is the current standard for 5-FU dosing but results in large interindividual difference of drug levels.
• •This is a multicenter, prospective German study assessing 5-FU AUC distributions in 434 patients receiving BSA-dosed 5-FU.
• •Only one-fifth of patients received 5-FU treatment within 5-FU AUC target range. The majority of patients were underdosed.
• •The only clinical parameter predicting dosing within target range was treatment setting (adjuvant versus metastatic).
• •Drug monitoring of 5-FU should be integrated into the clinical practice for optimized individualized cancer therapy.

Introduction

5-Fluorouracil (5-FU) is the backbone for the treatment of various tumors, including colorectal cancer (CRC), gastric cancer, pancreatic cancer, head and neck cancer (HNC), and breast cancer.¹ The current standard dosage method of 5-FU with body surface area (BSA) can be traced back to 1950s, a practice which causes significant differences in interindividual exposure.² Consequently, therapeutic failure could occur, such as reduced drug efficacy when patients are underdosed and severe toxicities when overdosing occurs. Pharmacokinetic measurement of 5-FU plasma concentration and its area under the curve (AUC), on the other hand, is a more precise method to assess systemic drug exposure and risk of toxicity. With a short half-life of ∼10-15 minutes, infusional 5-FU quickly reaches its steady state concentration (Css) and the AUC can be calculated as the product of Css and infusion time.² With a well-documented pharmacokinetic-pharmacodynamic correlation³ and a narrow therapeutic window, 5-FU is an ideal candidate for therapeutic drug monitoring (TDM) to guide personalized dose optimization.^1,2 A meta-analysis of a total of 654 patients with CRC or HNC from five clinical studies showed that TDM-guided 5-FU dosing resulted in a higher overall response rate and less toxicity compared with the standard BSA-based dosage method.⁴ Based on the clinical evidence, the International Association of Therapeutic Drug Monitoring and Clinical Toxicology (IATDMCT) strongly recommend 5-FU TDM in patients receiving 5-FU-containing regimens for CRC or HNC.² Regarding the target range of 5-FU AUC, Gamelin et al.⁵ initially proposed an optimal 5-FU AUC of 20-24 mg × h/l for a weekly 8-h continuous infusion of 5-FU monotherapy administered in patients with metastatic CRC. This target range is narrow with a width of only 4 mg × h/l, however, especially taking into account the intraindividual variability of ∼20%. Consequently, unnecessary dose adjustments may occur.² Based on the analysis of a large US dataset of 589 patients with CRC receiving FOLFOX6 (folinic acid, 5-fluorouracil, and oxaliplatin) therapy, Kaldate et al.⁶ proposed a wider therapeutic range of 20-30 mg × h/l, which reflects the greater tolerability of modern chemotherapy regimens and also realistically accommodates intraindividual variability and potential differences in the exposure-response relationships among different regimens.^2,6 The target range and dose adjustment algorithm proposed by Kaldate et al.⁶ were validated and shown to be feasible in the prospective, multicentre trial of Wilhelm et al.⁷ Nowadays, a target 5-FU AUC range of 20-30 mg × h/l is recommended as the therapeutic exposure range.²

Unfortunately, 5-FU TDM has not yet become part of standard care in Germany⁷ and many other countries due to various barriers: logistical issues including established sample collection and processing procedures, absence of ubiquitously available laboratory TDM tests, lack of support from oncology guidelines, and regulatory bodies and lack of training in treating physicians.^1,2 To underscore the importance of 5-FU TDM in cancer therapy with current real-world data, we conducted a multicenter, prospective observational study to determine the 5-FU AUC distributions in a large German cohort of patients receiving BSA-dosed 5-FU.

Methods

Patient inclusion

Patients with cancer aged ≥18 years who were treated with 5-FU-based chemotherapy were eligible for the study. Therapeutic setting could be metastatic or adjuvant. A total of 37 centers located throughout Germany participated in the study. The trial started in June 2017 and ended in January 2018. Patients’ clinical and demographic data were documented at baseline.

5-FU Administration, blood sampling, and plasma concentration determination

Continuous infusional 5-FU based on BSA-dosing was administered according to respective treatment regimens and treating physician’s decisions. Blood samples were collected from peripheral veins, drawn between 18 h after beginning of the infusion and before the end of the infusion, to ensure a steady state drug concentration with less variability.⁶ It was stipulated not to draw blood from the site where the central venous port system was located. In terms of handling of the samples, all participating centers were trained by the principal investigators to ensure standardized pretest conditions. Blood samples were stabilized using a ready-to-use dihydropyrimidine dehydrogenase (DPD) inhibitor gimeracil. Measurement of the 5-FU plasma concentration and determination of the AUC was carried out in central laboratories using the ‘My5-FU®’ assay, a competitive homogeneous two-reagent nanoparticle agglutination immunoassay (Saladax Biomedical, Inc.; Bethlehem, PA).

AUC measurement of 5-FU

The clinical outcome was measured by means of AUC. 5-FU AUC was determined to reflect the total drug exposure over a period of time, as calculated by multiplying Css of 5-FU by the total infusion time. Target range of 5-FU AUC was defined as 20-30 mg × h/l. Underdosing was defined when AUC was <20 mg × h/l, and overdosing was defined when AUC was >30 mg × h/l. AUC was described as a continuous variable, as well as a qualitative variable categorized in three groups (overdosing, target range, underdosing) or in two groups (target range reached or not).

Statistical analyses

The primary objective was to determine 5-FU AUC distributions in relation to the target range. The second objective was to investigate whether any clinical parameters correlate with achievement of the 5-FU target range.

Categorical or qualitative variables were described with counts and percentages. Quantitative continuous data were described with median and range (minimum, maximum), or mean and standard deviation (SD). All reported data were based on the latest information available in the database until 24 January 2018. Scatterplots were depicted for 5-FU AUC versus calculated relative dose per BSA (mg/m²) for total patients and for the subset of patients receiving FOLFOX therapy as well. The association between clinical parameters and achievement of 5-FU AUC in the target range was analyzed using bivariate analysis either by chi-square test, or Fisher’s exact test if the requirements for chi-square test were not fulfilled. To find potentially relevant covariates and factors to predict the achievement of the target range, univariate logistic regression was carried out as the first step. Parameters being significant at the 5% significance level would be included in a multivariate logistic regression model. Odds ratios were reported with 95% confidence interval (CI) to estimate the odds of 5-FU AUC in the target range.

A P value <0.05 was considered to be statistically significant. All statistical analyses were carried out using SAS software, release 9.4 (SAS Institute Inc., Cary, NC).

Ethics

This study was conducted in accordance with the Declaration of Helsinki, Good Clinical Practices, local ethical and legal requirements.

Study approval statement

This study was approved by the Ethics Committee II of Heidelberg University at the Medical Faculty Mannheim, approval number: 2016-559N-MA.

Study registration

The study was registered in Deutsches Register Klinischer Studien (DRKS, German Clinical Trials Register, https://www.drks.de) with the number DRKS00011486 on 21 December 2016 and with the German Cancer Society (Studybox) with the number ST-U041 on 25 July 2017.

Consent to participate statement

Written informed consents were obtained from all patients before participation in the study.

Results

Patient characteristics

A total of 454 case report forms (CRFs) were identified in the database of the trial. Of these, 20 CRFs were excluded (no results were available in nine cases, multiple measurements in seven cases, two cases with incomplete data entry, 1 duplicate CRF and one case with incorrect blood collection from the central venous port system). Finally, data of 434 patients were available for study analysis, as shown in Figure 1. The patients’ characteristics are summarized in Table 1. Their median age was 65 years (range, 34-89 years) and male patients represented a dominant proportion. The majority of patients (80.4%) had an Eastern Cooperative Oncology Group (ECOG) score of 0-1.

Figure 1.

CONSORT (Consolidated Standards of Reporting Trials) diagram.

CRF, case report form.

Table 1.

Patient characteristics at baseline

Characteristics	Number of patients (%) or value
Age at initial diagnosis, years
Median (range)	65 (34-89)
Sex
Female	147 (33.9)
Male	286 (65.9)
Missing	1 (0.2)
Body surface area (m2)
Mean (SD)	1.88 (0.20)
ECOG
0-1	349 (80.4)
2-3	23 (5.3)
Missing	62 (14.3)
Tumor location
Colorectal	309 (71.2)
Stomach	42 (9.7)
Pancreas	36 (8.3)
Esophagus	18 (4.1)
Other	29 (6.7)
Treatment type
Adjuvant	136 (31.3)
Metastatic	280 (64.5)
Missing	18 (4.2)

ECOG, Eastern Cooperative Oncology Group; SD, standard deviation.

Some 96.3% of tumors that required 5-FU treatment were located in the gastrointestinal tract. The most common tumor entity was CRC (71.2%), followed by stomach cancer (9.7%), pancreatic cancer (8.3%), and esophageal cancer (4.1%). Approximately two-thirds of the patients received 5-FU-based chemotherapy in the metastatic setting, whereas about one-third were treated in the adjuvant setting.

Treatment characteristics

Treatment characteristics are summarized in Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2023.101201. 5-FU was given to 28.1% of patients as monotherapy, otherwise as part of combination therapy such as FOLFOX, FOLFIRI (folinic acid, 5-fluorouracil, and irinotecan) or other regimens [e.g. FOLFOXIRI (folinic acid, 5-fluorouracil, irinotecan, and oxaliplatin), 5-FU/cisplatin, FLOT]. The median number of chemotherapy cycles was four, and one-fifth of the patients received the first cycle of 5-FU-containing chemotherapy during the study. Some 77.9% of patients were pretreated with 5-FU and 8.5% of patients already had dose reductions before study inclusion.

5-FU AUC distributions in relation to target range

Only 20.3% of all patients reached the target range of 5-FU AUC. The majority of patients (60.6%) were underdosed, whereas 19.1% of patients were overdosed. Scatterplots of AUC versus BSA dosage in all patients are shown in Figure 2A. Most patients achieving the target range had a dose of 5-FU between 1800 and 2600 mg/m². It is noteworthy that some patients were reported to have extremely high 5-FU AUC values. It is conceivable that in these patients blood samples were directly taken from the central venous port system. The scatterplot illustration for a subset of 144 patients receiving FOLFOX therapy is depicted in Figure 2. More patients treated with 2400 mg/m² 5-FU (standard dosage in FOLFOX6 or FOLFIRI regimen) were in the target range, whereas more patients receiving 1200 mg/m² (standard dosage in FOLFOX4 regimen) were underdosed (Figure 2B).

Figure 2.

Scatterplots of 5-FU AUC versus BSA dosage (A) in total patients and (B) in patients who received FOLFOX therapy. Purple stars represent 5-FU AUCs within the target range, olive green dots represent 5-FU AUCs under the target range (underdosing), and green triangles represent 5-FU AUCs over the target range (overdosing).

5-FU, 5-fluorouracil; AUC, area under the curve; BSA, body surface area.

Association between clinical characteristics and achievement of target range

Subgroups stratified by different patient or treatment characteristics with proportions of patients under, within, or over the target range of 5-FU AUC are illustrated in Figure 3. In the bivariate analysis for data without missing values, treatment type was the only clinical parameter that showed a significant association with achievement of target range (Supplementary Table S2, available at https://doi.org/10.1016/j.esmoop.2023.101201). Also in the univariate logistic regression, only treatment type significantly correlated with achievement of 5-FU AUC target range (Table 2). The odds of reaching 5-FU AUC target range in patients treated in a metastatic setting was 2.1 (95% CI 1.186-3.776, P = 0.011) times higher compared with patients receiving adjuvant therapy. A multivariate logistic regression was therefore not carried out.

Figure 3.

Subgroups stratified by clinical characteristics with proportions of patients within, under or over target range.

ECOG, Eastern Cooperative Oncology Group; FOLFIRI, chemotherapy regimen with folinic acid, 5-fluorouracil and irinotecan; FOLFOX, chemotherapy regimen with folinic acid, 5-fluorouracil and oxaliplatin.

Table 2.

Univariate logistic regression for clinical parameters and achievement of 5-FU AUC target range

Parameter	Parameter values	P value	OR estimates
Gender (N = 433)	Male versus female	0.8253	1.058 [0.643, 1.738]
Age (N = 430)	Continuous	0.2638	1.014 [0.990, 1.039]
ECOG (N = 372)	1 Versus 0	0.9628	0.920 [0.558, 1.517]
	2 Versus 0	0.9667	0.718 [0.231, 2.237]
	3 Versus 0	0.9637	Not estimable
BSA (N = 434)	Continuous	0.0596	0.310 [0.092, 1.048]
Location (N = 434)	Colorectal versus other	0.3645	0.793 [0.479, 1.310]
Treatment type (N = 416)	Metastatic versus adjuvant	0.0112	2.116 [1.186, 3.776]
Therapy (N = 434)	FOLFOX versus infusional FU	0.2645	1.897 [0.869, 4.143] 1.850 [0.970, 3.528] 1.194 [0.625, 2.283]
	FOLFIRI versus infusional FU	0.1930
	Other versus infusional FU	0.3618
Cycles (N = 420)	continuous	0.0791	1.024 [0.997, 1.051]
Dose reduction (N = 426)	No versus yes	0.3370	0.566 [0.234, 1.366]
	First cycle versus yes	0.3193	0.576 [0.269, 1.232]
Duration of infusion (N = 432)	Continuous	0.2725	0.993 [0.981, 1.005]

BSA, body surface area; ECOG, Eastern Cooperative Oncology Group; FU, 5-fluorouracil; OR, odds ratio.

Discussion

Our data including a total of 434 patients demonstrated that only 20.3% of measured 5-FU AUC values were within the target range of 20-30 mg × h/l, when the dosing method was based on BSA. This result is consistent with the evidence that the majority of 5-FU AUC was outside the target range in patients receiving BSA-dosed 5-FU.^5,7, 8, 9, 10

Some 19.1% of 5-FU AUC values were over the target range, revealing overdosing in those patients. Today, 5-FU overdosing can be reduced by determination of DPD status before therapy.¹ DPD is the key enzyme for metabolic degradation of FU which is encoded by the DPYD gene.¹ The half-life of 5-FU in patients with partial or complete DPD deficiency is markedly prolonged, which leads to drug accumulation and increased risk of severe and life-threatening toxicity, such as neutropenia, neurotoxicity, diarrhea, and stomatitis.¹¹ The European Medicines Agency therefore recommends a pretherapeutic DPD genotyping and/or phenotyping.¹¹ The DPD status of our study patients was not known, as its assessment was not a common practice by the time of study conduction and not required per study protocol. It is worth mentioning, however, that DPD mutations are not attributable to all DPD deficiencies or drug-related toxicities. The absence of DPYP gene variants does not rule out a DPD deficiency, and patients with normal DPD activities could still show elevated 5-FU plasma levels.³ For example, in the study of Johnson et al.,¹² only 43% of cancer patients experiencing moderate to severe toxicity during 5-FU treatment were identified as DPD deficient, whereas 57% of patients demonstrated normal DPD enzyme activity. In a further study reported by Bocci et al.,¹³ 3 of 188 patients with gastrointestinal tumors showed marked pharmacokinetics changes with prolonged half-time and low clearance, despite normal DPD enzymatic activity. The study of Schwab et al.¹⁴ also showed that non-genetic factors, such as gender, play a role in 5-FU metabolism. Additionally, a recent study has demonstrated that more patients with identified DPD deficiency who received an empirical dose reduction without TDM were underexposed than patients without DPD deficiency.¹⁵ Hence, in addition to the mandatory DPD testing before 5-FU-containing therapy, the measurement of 5-FU drug level and AUC can add extra value to monitor the systemic drug exposure and tailor the dose during the entire treatment course on an individual basis.

The majority of patients (60.9%) were below the target range of 5-FU AUC and therefore undertreated. This is in line with frequent underexposure observed in other studies.7, 8, 9, 10 A US study on 240 patients with CRC reported that BSA-based 5-FU dosing resulted in underdosing of almost half (48%) of the patients.⁸ Although oncologists often reduce drug dose in case of drug-related toxicities, they rarely increase the dose in the absence of intolerances, particularly when TDM is not available.² Compared with 5-FU overdosing with clinical warning signs, the underdosing issue remains rather undetectable and invisible. A lower 5-FU exposure, however, is linked with worse tumor response, higher recurrence rates, and inferior overall survival compared with higher drug exposure, as observed in patients with gastrointestinal cancers.16, 17, 18, 19 Underexposed patients are at risk of loss of drug efficacy, and TDM is the only way to solve this problem.²⁰

We also demonstrated that only the treatment setting (metastatic versus adjuvant) and no other clinical parameters are associated with achievement of the target range. Patients treated with 5-FU-containing chemotherapy in the metastatic setting had approximately two times higher odds to reach the therapeutic 5-FU AUC levels compared with those treated in the adjuvant setting. Our assumption is that adjuvant chemotherapy was more often given as the FOLFOX4 regimen with 5-FU administered at half the dose as that of the FOLFOX6 or FOLFIRI regimens (1200 mg/m² versus 2400 mg/m²), which are more often used for metastatic chemotherapy in our cohort. In the subgroup analysis of 144 patients receiving FOLFOX therapy, more patients receiving 1200 mg/m² 5-FU were underdosed compared with patients receiving 2400 mg/m² 5-FU. This is in line with US data reported by Saam and colleagues,¹⁰ that more patients with CRC on the FOLFOX4 regimen were under the target range of 5-FU AUC compared with those on the FOLFOX6 regimen (86% versus 51%). Apart from this, our study did not detect any clinical parameters influencing 5-FU AUC; that is to say, no certain subgroups would benefit more or less from the implementation of 5-FU drug monitoring based on our data.

TDM and pharmacokinetic-guided dosing for 5-FU is reported as feasible and beneficial in a multitude of studies.^6,7,10,21,22 It has been shown that 5-FU TDM is safe and can efficaciously increase the proportion of patients within the 5-FU AUC target range during an individual course of treatment.⁷ Moreover, a substantial benefit in efficacy parameters has been shown in a randomized French trial using 5-FU AUC instead of BSA dosing in patients receiving 5-FU monotherapy.⁵ Also, from the perspectives of pharmacoeconomics, 5-FU TDM for metastatic CRC and HNC is cost-effective both in the US and UK health care system, as the pay per quality-adjusted life-year was much lower than a threshold of $50 000.23, 24, 25 This cost-effectiveness is probably driven by reducing adverse events and improving drug efficacy through TDM.^3,25 To foster the implementation of TDM into oncology clinical practice, inclusion of TDM into clinical guidelines and covering logistical costs might be effective incentives and strategies.²

Our study has limitations. First, this study is designed as a cross-sectional study but not a longitudinal study, so that toxicity and efficacy as follow-up data were not available. Also, dose adjustment as a subsequent intervention was beyond the study design due to its non-interventional nature. Second, there are some wide outliers of AUC which may be caused by preanalytical errors. Despite the standardization of pre-analytic procedures per study protocol, a calculated AUC of 0 is probably because the 5-FU infusion pump was empty when blood was drawn or that the stabilizer was not immediately put into the blood sample, and an extremely high AUC could be explained by an accidental draw from the central venous catheter. We decided to keep all measured values to avoid selection bias.

Nevertheless, the present study demonstrated with recent real-world data from a large cohort again a great interindividual variability of systematic drug exposure measured by 5-FU AUC when using the BSA dosing method. It further confirms observations from other studies that most patients are given suboptimal doses. We aim to raise physicians’ and regulators’ awareness to carry out drug monitoring when treating patients with 5-FU.

Conclusion

BSA-based 5-FU dosing resulted in a wide variability in 5-FU AUC of patients with gastrointestinal tumors. The majority of patients were given suboptimal 5-FU doses. Patients treated in the metastatic setting had higher odds of reaching the 5-FU target range compared with those treated in the adjuvant setting. 5-FU TDM is indispensable to tailor 5-FU dose on an individual basis and should be integrated into the clinical practice.