Influence of CYP3A4/5 and ABC transporter polymorphisms on lenvatinib plasma trough concentrations in Japanese patients with thyroid cancer

Lenvatinib is a substrate of cytochrome P450 (CYP) 3A and ATP-binding cassette (ABC) transporters. In this study, we aimed to evaluate how CYP3A4/5 and ABC transporter polymorphisms affected the mean steady-state dose-adjusted plasma trough concentrations (C0) of lenvatinib in a cohort of 40 Japanese patients with thyroid cancer. CYP3A4 20230G > A (*1G), CYP3A5 6986A > G (*3), ABCB1 1236C > T, ABCB1 2677G > T/A, ABCB1 3435C > T, ABCC2 −24C > T, and ABCG2 421C > A genotypes were determined using polymerase chain reaction-restriction fragment length polymorphism. In univariate analysis, there were no significant differences in the mean dose-adjusted C0 values of lenvatinib between the ABCB1, ABCG2, and CYP3A5 genotypes. However, the mean dose-adjusted C0 values of lenvatinib in patients with the CYP3A4*1/*1 genotype and ABCC2 −24T allele were significantly higher than those in patients with the CYP3A4*1G allele and −24C/C genotype, respectively (P = 0.018 and 0.036, respectively). In multivariate analysis, CYP3A4 genotype and total bilirubin were independent factors influencing the dose-adjusted C0 of lenvatinib (P = 0.010 and 0.046, respectively). No significant differences were found in the incidence rates of hypertension, proteinuria, and hand-foot syndrome following treatment with lenvatinib between the genotypes of CYP3A4/5 and ABC transporters. Lenvatinib pharmacokinetics were significantly influenced by the CYP3A4*1G polymorphism. If the target plasma concentration of lenvatinib for efficacy or toxicity is determined, elucidation of the details of the CYP3A4*1G genotype may facilitate decision-making related to the appropriate initial lenvatinib dosage to achieve optimal plasma concentrations.

In addition, in humans, three ABC subfamilies, B, C, and G, contain efflux transporters for a large number of drugs. Among these ABC subfamilies, ABCB1, ABCG2, and ABCC2 (multidrug resistance-associated protein [MRP] 2) are important for the drug transporters. To date, however, in both in vitro and in vivo studies, the influence of MRP2 on membrane transport of lenvatinib has not been reported. In addition to P-glycoprotein and BCRP, MRP2 may also play a role in lenvatinib pharmacokinetics. In addition, no studies have reported the effects of ABCC2 polymorphisms on lenvatinib plasma concentrations. Therefore, in the present study, we investigated the effects of polymorphisms in CYP3A4 20230G > A, CYP3A5 6986A > G, ABCB1 1236C > T, ABCB1 2677G > T/A, ABCB1 3435C > T, ABCG2 421C > A, and ABCC2 −24C > T on the mean steady-state dose-adjusted plasma trough concentration (C 0 ) of lenvatinib in 40 patients with thyroid cancer.

Methods
Patients and protocols. Forty Japanese patients with thyroid cancer (27 women and 13 men) taking lenvatinib (LENVIMA; Eisai Co., Ltd., Tokyo, Japan) were consecutively enrolled in this study. All patients were treated at Ito Hospital between June 2015 and April 2018. The eligibility criteria were as follows: adequate bone marrow function (neutrophil count of 1500/μL or more, platelet count of 100,000/μL or more, and haemoglobin level of 8.0 g/dL or more), renal function (serum creatinine levels less than 2.0 mg/dL), and hepatic function (total bilirubin level less than 2.0 mg/dL, aspartate transferase and alanine transferase levels less than or equal to 2.5 times the upper limit of the normal range). Table 1 lists the clinical characteristics of patients prior to lenvatinib therapy. Patients were excluded if they had taken drugs that affected CYP3A, P-glycoprotein, or BCRP function. Lenvatinib was administered as medical care for patients with thyroid cancer according to a Drug Interview Form 8 . The study was approved by the Ethics Committee of Ito Hospital and Akita University School of Medicine and was carried out according to the recommendations outlined in the 1964 Declaration of Helsinki and its later amendments. All patients who participated in this study provided written informed consent prior to enrolment.
Oral lenvatinib administration was carried out at dosages of 4-24 mg once daily. The daily dose of lenvatinib was reduced as needed based on the grade of each side effect, as determined by the Common Terminology Criteria for Adverse Events version 4.0. Intake of foods or drugs known to alter CYP3A, P-glycoprotein, or BCRP function was not permitted during the treatment period (withdrawal criteria). The plasma concentrations of lenvatinib were monitored 24 ± 4 h after administration, and trough plasma concentrations (C 0 ) were evaluated every 2 weeks for the first 3 months after initiation of lenvatinib therapy and then every 1 month thereafter. Assays of lenvatinib C 0 were performed monthly until lenvatinib treatment discontinuation (unacceptable toxicity, disease progression, or death of patient) or for a maximum of 2 years (maximum 27 sampling points). Blood samples except the sampling time of 24 ± 4 h after administration and the steady-state was excluded from analysis. In addition, plasma samples from patients who developed serious hepatic or renal dysfunction after initiation of lenvatinib therapy were excluded from analysis because of the influence on lenvatinib pharmacokinetics (Fig. 1). Plasma was isolated by centrifugation at 1,900 × g for 15 min and was stored at −40 °C until analysis.
Pharmacokinetics analysis. Plasma concentrations of lenvatinib were measured by high-performance liquid chromatography (HPLC), as has been used previously for quantitative analysis of gefitinib plasma concentrations 13 . Analysis of the calibration curve for lenvatinib in human plasma showed that the curve was linear  www.nature.com/scientificreports www.nature.com/scientificreports/ for the concentration range from 5 to 1000 ng/mL. Moreover, for this assay, the limit of quantification (LOQ) of lenvatinib was 5 ng/mL. For the concentration range from 5 to 1000 ng/mL, the coefficients of variation (CVs) were lower than 12.6%, and the accuracies of the intra-and interday assays were within 10.6%.
Pharmacogenomics analysis. Extraction of DNA from peripheral blood samples was carried out using a QIAamp Blood Mini Kit (Qiagen, Tokyo, Japan). Genotypes CYP3A4 20230G > A (*1G) (rs2242480), within intron 10 of the human CYP3A4 gene, and CYP3A5 6986A > G (*3) (rs776746) were identified using polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) 14 . Genotyping procedures identifying the C and T alleles in exon 12 (1236C > T, rs1128503), the G and T/A alleles in exon 21 (2677G > T/A, rs2032582), and the C and T alleles in exon 26 (3435C > T, rs1045642) of the ABCB1 gene were performed using PCR-RFLP as previously reported [15][16][17] . PCR-RFLP was used for genotyping of the ABCG2 421C > A (rs2231142) polymorphism, as described by Kobayashi et al. 18 . Genotyping procedures identifying the C and T alleles in exon 1 (−24C > T, rs717620) of the ABCC2 gene were carried out using PCR-RFLP, as described by Naesens et al. 19 . All frequencies for the different analysed loci were at Hardy-Weinberg equilibrium.
Statistical analyses. The Kolmogorov-Smirnov test was applied to assess the distribution in each data set.
The results used the average of dose-adjusted C 0 values of lenvatinib during therapy for one patient. Dose-adjusted C 0 values for each genotype were presented as medians (minimum-maximum). To elucidate differences between these groups, we applied Kruskal-Wallis tests or Mann-Whitney U tests. Correlations in continuous variables between groups were determined using Spearman's rank correlation coefficient test, with results presented as correlation coefficients (r). The effects of factors in univariate analysis were evaluated using stepwise multiple linear regression analysis. For each patient, dummy variables were used to replace the genotypes of CYP3A4/5 and ABC transporters (1 and 0 in 2 groups; 1 and 0, 0 and 0, and 0 and 1 in 3 groups). Chi-square tests were used to examine differences in categorical data. For post-hoc power analysis, the effect size calculated in comparison with the lenvatinib dose-adjusted C 0 between CYP3A4 or ABCC2 genotypes showed significant differences. An effect size of greater than 0.5 was considered clinically meaningful 20 . Power was calculated using G*Power ver. 3.1 software. Results with p values of less than 0.05 were considered significant, and SPSS 20.0 for Windows (SPSS IBM Japan Inc., Tokyo, Japan) was used for all statistical analyses.
There were no significant differences in the mean dose-adjusted C 0 of lenvatinib between genotypes for the ABCB1, ABCG2, and CYP3A5 polymorphisms, although the dose-adjusted C 0 of lenvatinib in patients with the ABCG2 421A allele or CYP3A5*3/*3 tended to be higher than those with other corresponding genotypes (Table 2). However, the mean dose-adjusted C 0 of lenvatinib in patients with the CYP3A4*1/*1 genotype was significantly higher than that in patients with the CYP3A4*1G allele (P = 0.018, effect size = 0.863; Table 2). In addition, the mean dose-adjusted C 0 of lenvatinib in patients with the ABCC2 −24T allele was significantly higher than that in patients with −24C/C genotypes (P = 0.036, effect size = 0.605; Table 2). There were also significant correlations with total bilirubin at lenvatinib initiation and the mean dose-adjusted C 0 of lenvatinib at the steady state (P = 0.037; Table 2). The dose-adjusted C 0 of lenvatinib in patients with both the CYP3A4*1/*1 and ABCC2 −24T alleles (median 6.70 ng/mL/mg) was about 1.5-fold higher than that in patients with both the CYP3A4*1G/*1G and ABCC2 −24C/C genotypes (median 4.42 ng/mL/mg; P = 0.007; Fig. 2).  Table 3. The CYP3A4 genotype (CYP3A4*1/*1) and total bilirubin were independent factors influencing the dose-adjusted C 0 of lenvatinib (P = 0.010 and 0.046, respectively); however, the determination coefficient for the dose-adjusted C 0 of lenvatinib was 0.226. No significant differences were found in the incidence rates of hypertension, proteinuria, and hand-foot syndrome following treatment with lenvatinib between the genotypes of CYP3A4/5 and ABC transporters (Table 4).

Discussion
To the best of our knowledge, this is the first study to report the effects of CYP3A4/5 and ABC transporter polymorphisms on lenvatinib pharmacokinetics in patients with thyroid cancer. Our findings showed that lenvatinib pharmacokinetics were significantly influenced by the CYP3A4 20230G > A polymorphism; however, the ABCB1, ABCG2, and CYP3A5 polymorphisms had no effect. Notably, univariate analysis showed that the dose-adjusted C 0 of lenvatinib was significantly higher in patients having the ABCC2 −24T allele than in patients having the −24C/C genotype (P = 0.036). Accordingly, in patients with both the CYP3A4*1G/*1G and ABCC2 −24C/C genotypes, the dose-adjusted C 0 of lenvatinib was significantly lower than that in patients with other genotypes. However, the appearances of hypertension, proteinuria, and hand-foot syndrome following lenvatinib treatment appeared to be unrelated to these genetic polymorphisms. www.nature.com/scientificreports www.nature.com/scientificreports/ Eighty-eight percent of cases of the CYP3A4*1G polymorphism 20230G > A have been reported to be linked to CYP3A5*1 21 . Notably, in a previous study of patients harboring the CYP3A4*1G allele, metabolic activity for the substrate of CYP3A4 was increased; that is, patients harboring the CYP3A4*1G allele showed significantly lower dose-adjusted C 0 values of tacrolimus than patients harboring the CYP3A4*1/*1 genotype 14 . As previously demonstrated 21 , patients with the CYP3A4*1G allele exhibited significantly lower dose-adjusted C 0 values for lenvatinib than patients having the CYP3A4*1/*1 genotype. However, the dose-adjusted C 0 values of lenvatinib did not differ significantly between the 2 CYP3A5 genotypes. Twelve percent of CYP3A4*1G polymorphisms that are not linked to CYP3A5*1 seem to show increased correlations between CYP3A4*1G polymorphisms and lenvatinib pharmacokinetics.
In an in vitro study, the ABCC2 −24T variant was reported to be associated with decreased transport activity of MRP2 22 . Similarly, in the present study, the dose-adjusted C 0 of lenvatinib was also significantly increased in patients having the ABCC2 −24T allele compared with that in patients having the −24C/C genotype. Therefore, ABCC2 polymorphism may also contribute to interindividual differences in lenvatinib pharmacokinetics. However, the SNP in position 421C > A of ABCG2 is the most prevalent allele in Japanese individuals (33%) and exhibits functional importance 23 . The level and function of ABCG2 protein expressed from the 421A allele are reduced compared with those of the 421C/C protein 18 . To date, the plasma concentrations of other TKIs, particularly imatinib, in patients with the ABCG2 421A allele have been reported to be higher than those in patients with the 421C/C genotype 24,25 . Notably, our current findings demonstrated that in patients having the ABCG2 421A allele, the dose-adjusted C 0 of lenvatinib tended to be higher than in patients having the 421C/C genotype, although this difference was not significant. Furthermore, each polymorphism in the human ABCB1 gene (1236C > T, 2677G > T/A, and 3435C > T) has been associated with altered expression and function of P-glycoprotein in various human tissues [15][16][17] . The dose-adjusted C 0 of lenvatinib in patients with the ABCB1 1236 T/T or 3435 T/T genotype also tended to be higher than that in patients with the corresponding 1236C/C or 3435C/C genotype. Accordingly, further studies with larger sample sizes are needed; however, in addition to   Table 3.
Stepwise multiple regression analysis of explanatory variables for the dose-adjusted C 0 of lenvatinibTable 3. Stepwise multiple regression analysis of explanatory variables for the dose-adjusted C 0 of lenvatinib. C 0 , predose concentration. SE, standard error; SRC, standardized regression coefficient. (2019) 9:5404 | https://doi.org/10.1038/s41598-019-41820-y www.nature.com/scientificreports www.nature.com/scientificreports/ P-glycoprotein and BCRP, MRP2 may also contribute to the absorption and disposition of lenvatinib. MRP2 is localized at the apical membrane of enterocytes in the intestine and plays a key role in limiting the absorption of substrate drugs incorporated orally. In the liver, MRP2 localizes to the canalicular membrane of hepatocytes and mediates the efflux of substrate drugs into the bile 26 .
Our current results showed that patients having both the CYP3A4*1/*1 and ABCC2 −24T alleles administered the same daily dose of lenvatinib exhibited lenvatinib C 0 values about 1.5-fold higher than those in patients having both the CYP3A4*1G/*1G and ABCC2 −24C/C genotypes. Higher lenvatinib plasma concentrations may be associated with higher rates of some adverse events. In lenvatinib therapy, evaluation of lenvatinib plasma concentrations may be necessary to provide individual treatment through dose adjustment because of improved efficacy or to avoid adverse events. Identification of therapeutic target plasma concentrations indicating the relationships of drug exposure-efficacy or toxicity will then be required. When target plasma concentrations of lenvatinib are elucidated, the present results may be useful clinically. Specifically, elucidation of information regarding the primary CYP3A4 genotype and minor ABCC2 genotype prior to initiation of lenvatinib therapy may facilitate decision-making related to the appropriate daily dosage for obtaining optimal lenvatinib plasma exposure. Further studies are necessary in the future to examine this possibility.
Our study had several limitations. First, the number of patients treated with lenvatinib therapy was only 40, which may have hindered the precise determination of lenvatinib pharmacogenomics. Second, because some polymorphisms were of low frequency, the associations observed may have been accidental. Therefore, further studies with larger patient cohorts are needed to confirm our results. However, we investigated the validity of the sample size from the results obtained from the study and found that the effect sizes in comparison with the lenvatinib dose-adjusted C 0 values between CYP3A4 or ABCC2 genotypes were 0.863 and 0.605, respectively, supporting the clinical significance of the results.

Conclusion
Overall, our results demonstrated that the mean steady-state dose-adjusted C 0 of lenvatinib was significantly affected by CYP3A4 polymorphism but not ABCB1, ABCG2, or CYP3A5 polymorphisms. In addition to P-glycoprotein and BCRP, MRP2 may also contribute to the absorption and disposition of lenvatinib. If the target plasma concentration of lenvatinib for efficacy or toxicity is determined, elucidation of details of the CYP3A4*1G genotype may facilitate decision-making related to the appropriate initial lenvatinib dosage to achieve optimal plasma concentrations.