Background

Gefitinib is a selective inhibitor of the epidermal growth factor receptor (EGFR) and is used for the treatment of patients with non-small cell lung cancer (NSCLC) [1]. Gefitinib is metabolised by cytochrome P450 (CYP) 2D6 to the O-desmethyl metabolite [2–5], which inhibits subcellular EGFR tyrosine kinase through a mechanism similar to that of gefitinib (half-maximal inhibitory concentration [IC50] = 0.036 versus 0.022 µM, respectively). However, because the transition to tissue is low for this metabolite, O-desmethyl gefitinib has minor effects on tumour growth [6]. In an alternative pathway, gefitinib is metabolised to its morpholine ring-opened metabolite by CYP3A4 and CYP3A5 [2]. Among gefitinib metabolites, only O-desmethyl gefitinib can be identified in human plasma at concentrations similar to those of the parent compound; the plasma concentrations of other gefitinib metabolites are very low or even undetectable [2, 4, 7].

The formation of O-desmethyl gefitinib seems to be dependent on CYP2D6 activity [8–10]. Therefore, polymorphisms in the CYP2D6 gene influence the rate of elimination of gefitinib. Single administration of gefitinib has been reported to yield concentrations that are about twofold higher in poor metabolisers (PMs) of CYP2D6 (e.g. *4/*4, *4/*5, and *3/*4) compared with extensive metabolisers of CYP2D6 (patients having the *1 or *2 allele) [9]. In Asian populations, however, the non-functional CYP2D6*3, *4, and *6 allelic variants have not been observed, and the frequency of the non-functional CYP2D6*5 allele is very low [11]. The CYP2D6*10 allele, the most common allele in Asians at frequencies of 33–50 %, is a reduced-function allele [12, 13]. However, in our previous study, Japanese patients with NSCLC having the CYP2D6*5 or *10 allele did not exhibit abnormalities in gefitinib exposure [14].

The effects of CYP2D6 polymorphisms on gefitinib exposure have been studied [9, 14]; however, the effects of CYP2D6*5 and *10 polymorphisms on the formation of O-desmethyl gefitinib from gefitinib have not been clarified. Because in vitro studies have reported that the formation of O-desmethyl gefitinib from gefitinib is mediated by CYP2D6 alone [3–5], individual variability in CYP2D6 metabolism could be more relevant for O-desmethyl gefitinib than for the parent compound. Accordingly, in patients with NSCLC, we may be able to estimate the involvement of CYP2D6 polymorphisms in gefitinib exposure by using the individual pharmacokinetics of gefitinib and O-desmethyl gefitinib. In addition, the relationship between O-desmethyl gefitinib exposure and side effects, such as diarrhoea, skin rash, and hepatotoxicity, has not been clarified, although this metabolite is not expected to affect the tumour [6]. O-Desmethyl gefitinib is excreted predominantly in faeces [2, 15]. ATP-binding cassette (ABC) transporters, such as P-glycoprotein (encoded by the ABCB1 gene) and breast cancer resistance protein (BCRP, encoded by the ABCG2 gene), may be involved in the biliary excretion of O-desmethyl gefitinib because gefitinib is a substrate of P-glycoprotein and BCRP [7, 16]. However, no studies have reported the effects of ABCB1 and ABCG2 polymorphisms on O-desmethyl gefitinib exposure.

In the present study, we investigated the effects of polymorphisms in CYP2D6, ABCB1, and ABCG2 and the side effects induced by gefitinib on the pharmacokinetics of O-desmethyl gefitinib.

Methods

Patients and protocols

Thirty-six Japanese patients with NSCLC (24 women and 12 men) taking gefitinib (Iressa®; AstraZeneca, Osaka, Japan), who were treated at the Akita University Hospital from January 2011 through April 2015, were prospectively enrolled in the study. All 28 patients with NSCLC who had undergone analysis of the CYP2D6 genotype in our previous study [14] were included in the current study. The demographic and clinical characteristics of the patients on day 14 after gefitinib therapy are listed in Table 1. The study protocol was approved by the Ethics Committee of Akita University School of Medicine (number 1140), and all patients provided written informed consent for participation in the study. The toxicity grades for diarrhoea, skin rash, and hepatotoxicity were categorised as described in the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0.

Table 1 Clinical characteristics of patients receiving gefitinib
Full size table

Gefitinib 250 mg was orally administered once daily at 08:00 (morning time). During the study period, the patients were not permitted to consume or ingest drugs or foods that were known to affect CYP3A and P-glycoprotein function. On day 14 after beginning gefitinib therapy, whole blood samples were collected just prior to and at 1, 2, 4, 6, 8, 12, and 24 h after oral gefitinib administration. Plasma was isolated by centrifugation at 1900×g for 15 min and was stored at −40 °C until analysis.

Analytical methods

O-Desmethyl gefitinib was purchased from Toronto Research Chemicals Inc. (Ontario, Canada). Plasma concentrations of gefitinib and O-desmethyl gefitinib were measured by high-performance liquid chromatography (HPLC) [17]. Namely, following the addition of erlotinib (25 ng/10 µL methanol) as an internal standard to a 100-µL plasma sample, the plasma sample was diluted with 900 μL water and vortexed for 30 s. This mixture was applied to an Oasis HLB extraction cartridge that had been activated previously with methanol and water (1.0 mL each). The cartridge was then washed with 1.0 mL water and 1.0 mL of 40 % methanol in water and eluted with 1.0 mL of 100 % methanol. Eluates were evaporated to dryness in a vacuum at 40 °C using a rotary evaporator (Iwaki, Tokyo, Japan). The resulting residue was then dissolved in 20 μL methanol and vortexed for 30 s; 20 μL of the mobile phase was added to the sample, and the sample was vortexed for another 30 s. A 20-μL aliquot of the sample was then processed by HPLC. The HPLC system was comprised of a PU-2080 plus chromatography pump (JASCO, Tokyo, Japan) equipped with a CAPCELL PAK C18 MG П HPLC column (250 mm × 4.6 mm I.D.; Shiseido, Tokyo, Japan), a UV-2075 light source, and an ultraviolet detector (JASCO). The mobile phase was 0.5 % KH2PO4 (pH 3.5)–acetonitrile–methanol (55:25:20, v/v/v) and was degassed in an ultrasonic bath prior to use. The flow rate was 0.5 mL/min at ambient temperature, and sample detection was carried out at 250 nm. The coefficients of variation for intra- and interday assays were less than 8.4 %. Accuracies for intra- and interday assays were within 3.1 %. The limits of quantification for gefitinib and O-desmethyl gefitinib were each 10 ng/mL.

Identification of genotypes

DNA was extracted from peripheral blood samples using a QIAamp Blood Mini Kit (Qiagen, Tokyo, Japan) and was stored at −80 °C until analysis. The CYP2D6*5 (deleted) allele and CYP2D6*10 (reduced) allele were identified using long polymerase chain reaction (PCR) analysis and the PCR–restriction fragment length polymorphism (RFLP) method, respectively, as described by Naveen et al. [18]. The patients were grouped into three groups according to the combination of alleles: extensive metabolisers (EMs; CYP2D6*1/*1, *1/*2, and *2/*2; n = 13), heterozygous EMs (CYP2D6*1/*5, *2/*5, *1/*10, and *2/*10; n = 18), and intermediate metabolisers (IMs; CYP2D6*5/*10 and *10/*10; n = 5; Table 1). Genotyping procedures identifying the C and T alleles in exon 26 (3435C > T) of the ABCB1 gene were performed using the PCR–RFLP method described by Cascorbi et al. [19]. The ABCG2 421C > A polymorphism was genotyped by the PCR–RFLP method described by Kobayashi et al. [20]. The analytic results obtained from PCR–RFLP were confirmed using a fully automated single nucleotide polymorphism (SNP) detection system (prototype i-densy™, ARKRAY Inc., Kyoto, Japan). All frequencies for the different analysed loci were at Hardy–Weinberg equilibrium.

Pharmacokinetic analysis

The pharmacokinetic analyses of gefitinib and O-desmethyl gefitinib were carried out using standard non-compartmental methods with Phoenix WinNonlin Version 6.4 (Pharsight Co., Mountain View, CA, USA). The area under the observed plasma concentration–time curve (AUC) from 0 to 24 h was calculated using the linear trapezoidal rule.

Statistical analyses

The Shapiro–Wilk test was used to assess the data distribution. The clinical characteristics of patients taking gefitinib were expressed as the number or mean value (two-sided 95 % confidence interval) on day 14 after beginning gefitinib therapy. The AUC0–24 of O-desmethyl gefitinib, gefitinib, total AUC0–24 of gefitinib plus O-desmethyl gefitinib, and AUC ratio of O-desmethyl gefitinib to gefitinib were expressed as medians and first to third quartiles. Kruskal–Wallis tests or Mann–Whitney U tests were used to determine the differences in continuous values between groups. Spearman’s rank correlation coefficient tests were applied to assess correlations between the AUC value and patient clinical characteristics, and all results were expressed as the correlation coefficient of determinant (r). A stepwise multiple linear regression analysis for AUC value was performed to determine the effects of factors with P values of greater than 0.2 as examined in univariate analysis. For each patient, the genotypes of the CYP2D6 or drug transporters were replaced with dummy variables (1 and 0, 0 and 1, or 0 and 0). Differences or correlations with P values of less than 0.05 were considered statistically significant. Statistical analysis was performed using SPSS 20.0 for Windows (SPSS IBM Japan Inc., Tokyo, Japan).

Results

There were no significant correlations between the AUC0–24 of O-desmethyl gefitinib or the AUC ratio of O-desmethyl gefitinib to gefitinib and gender, age, body weight, body surface area, or laboratory test values (Table 2).

Table 2 Comparisons and correlations of clinical characteristics and AUC0–24 of O-desmethyl gefitinib or AUC ratio of O-desmethyl gefitinib to gefitinib
Full size table

The genotype frequency for the CYP2D6 polymorphism in 36 Japanese patients with NSCLC is shown in Table 1. The mean plasma concentrations of O-desmethyl gefitinib were lowest in patients having CYP2D6*5/*10 or *10/*10, intermediate in the those having heterozygous CYP2D6*1 or *2 alleles, and highest in those having CYP2D6*1/*1, *1/*2, or *2/*2 (Fig. 1). The median AUC0–24 of O-desmethyl gefitinib in patients with CYP2D6*5/*10 or *10/*10 was 1460 ng h/mL, whereas that in patients having CYP2D6*1/*1, *1/*2, or *2/*2 was 12,523 ng h/mL (P = 0.021, Table 3). The ratio of the AUC of O-desmethyl gefitinib to that of gefitinib differed significantly among the three groups, with the median ratio for homozygous EMs to heterozygous EMs to IMs being 1.41:0.86:0.24 (P = 0.030, Table 3). However, there was no significant difference in the AUC0–24 of gefitinib among the CYP2D6 genotypes (P = 0.467). On the other hand, there were no significant differences in the AUC0–24 values of O-desmethyl gefitinib in patients with the ABCB1 or ABCG2 transporter genotypes (Table 3).

Fig. 1
figure 1

Mean ± SD plasma concentration–time profiles of gefitinib (open circles) and O-desmethyl gefitinib (solid circles). a Homozygous extensive metabolisers (CYP2D6*1/*1, *1/*2, and *2/*2, n = 13), b heterozygous extensive metabolisers (CYP2D6*1/*5, *2/*5, *1/*10, and *2/*10, n = 18), and c intermediate metabolisers (IMs; CYP2D6*5/*10, and *10/*10, n = 5)

Full size image
Table 3 Comparison of the AUC0–24 of O-desmethyl gefitinib and AUC ratio of O-desmethyl gefitinib to gefitinib between CYP or drug-transporter genotype groups
Full size table

The results of multivariate analysis including covariates are listed in Table 4. CYP2D6 EMs, such as those having *1/*1, *1/*2, and *2/*2 (P = 0.012), were independently predictive for higher AUC0–24 of O-desmethyl gefitinib. In addition, for the AUC ratio of O-desmethyl gefitinib to gefitinib, CYP2D6*1/*1, *1/*2, and *2/*2 (P = 0.001) and younger age (P = 0.032) were independently predictive.

Table 4 Stepwise selection multiple linear regression analysis of explanatory variables for the AUC0–24 of O-desmethyl gefitinib and O-desmethyl gefitinib to gefitinib
Full size table

The AUC0–24 values of O-desmethyl gefitinib in patients with diarrhoea (n = 17) were significantly lower than those in patients without diarrhoea (n = 19), suggesting that exposure to gefitinib, but not its O-desmethyl metabolite, contributed to the incidence of diarrhoea (Table 5). In addition, there was no significant difference in O-desmethyl gefitinib exposure between patients with (n = 22) and without (n = 14) a skin rash and between patients with (n = 20) and without (n = 16) hepatotoxicity (Table 5).

Table 5 Comparisons of side effects and AUC0–24 values of O-desmethyl gefitinib
Full size table

Discussion

Side effects of diarrhoea, skin rash, and hepatotoxicity induced by gefitinib were unrelated to the plasma exposure of the active O-desmethyl metabolite and the total exposure of gefitinib and O-desmethyl gefitinib. The patients with higher exposure to the parent compound gefitinib exhibit initial onset of diarrhoea, followed later by hepatotoxicity with continuous administration [14, 17]. Higher gefitinib exposure, but not O-desmethyl metabolite exposure, seemed to be associated with the incidence of diarrhoea and hepatotoxicity. On the other hand, in patients showing lower plasma exposure of O-desmethyl gefitinib, it is possible that the concentration of the metabolite may be high in the gastrointestinal tract, potentially stimulating the gut and resulting in induction of diarrhoea, because O-desmethyl gefitinib is excreted predominantly in faeces after metabolism in the liver [2, 15]. However, three of five patients with the CYP2D6*5/*10 or *10/*10 genotype, showing low metabolite formation, also developed diarrhoea. Therefore, the analysis of the plasma concentrations of gefitinib, but not O-desmethyl gefitinib, may be necessary because of its direct relationship with the occurrence of diarrhoea and hepatotoxicity. On the other hand, gefitinib-induced skin rash did not appear to be related to exposure of either gefitinib or O-desmethyl gefitinib.

CYP2D6*5 and *10 polymorphisms were associated with the formation of O-desmethyl gefitinib from gefitinib. By analysis of the O-desmethyl metabolite of gefitinib, we were able to confirm the effects of the CYP2D6 polymorphism on the formation of O-desmethyl gefitinib. In one clinical study, single administration of gefitinib in healthy volunteers resulted in a mean AUC0–∞ of O-desmethyl gefitinib of 1640 ng h/mL in CYP2D6 EMs; however, no such value was found in CYP2D6 PMs, i.e. those with genotypes *4/*4, *4/*5, and *3/*4 [9]. This previous result [9] was similar to the results obtained from our present study. However, the mean AUC0–∞ of gefitinib in CYP2D6 PMs was about twofold higher than that in CYP2D6 EMs with the *1 or *2 allele (3060 vs. 1430 ng h/mL) [9]. In the present study, there were no significant differences in the AUC0–24 of gefitinib between CYP2D6 EMs and IMs. We cannot completely explain this discrepancy. CYP3A4 and CYP3A5 are the main enzymes involved in other gefitinib metabolic pathways [4, 5]. Repeated administration of gefitinib in patients with reduced CYP2D6 activity may result in metabolism of gefitinib mainly through the CYP3A4 and CYP3A5 pathways. van Waterschoot et al. [21] reported that gefitinib is a potent stimulator of triazolam metabolism by CYP3A4 both in vitro and in vivo. The higher gefitinib exposure after single administration of gefitinib in CYP2D6 IMs or PMs may enhance CYP3A4 activity to a greater extent than that in CYP2D6 EMs by repetitive administration of gefitinib. Consequently, no significant differences in the AUC0–24 of gefitinib at steady state may be observed between CYP2D6 EMs and IMs.

To date, more than 100 allelic variants have been reported for CYP2D6 [11]. However, in our present study, only CYP2D6*5 and *10 were analysed. Therefore, the CYP2D6 allele having no or reduced enzyme activity may be included in the CYP2D6 EMs classified in the present study. In fact, the AUC0–24 of O-desmethyl gefitinib and the AUC ratio of O-desmethyl gefitinib to gefitinib in patients with CYP2D6*1/*1, *1/*2, or *2/*2 in the present study exhibited large ranges between 1208 and 46,320 ng h/mL and between 0.07 and 5.50, respectively. There was about a 38-fold variation in patients with CYP2D6*1/*1, *1/*2, or *2/*2 between the lowest and highest measurements of O-desmethyl gefitinib formation from gefitinib. In the Asia population, by analysing other CYP2D6 variants, such as *41 and *49 [11], our results may be more clear. In the previous study by Swaisland et al. [9], there was about a 39-fold variation between the lowest and highest gefitinib AUC0-∞ values in CYP2D6 PMs, showing that there was considerable overlap in the ranges of individual values. Therefore, they concluded that CYP2D6 genotyping prior to initiation of gefitinib therapy is not necessary [9]. We agree with this conclusion and recommend that plasma gefitinib concentrations should be analysed after beginning gefitinib therapy rather than analysing CYP2D6 polymorphisms before initiating therapy because of the direct relationship of plasma gefitinib concentrations with side effects [14, 17]. On the other hand, however, Xin et al. [22] reported that the trough plasma concentration of gefitinib is not significantly associated with diarrhoea and hepatotoxicity. Hence, additionally studies having larger sample sizes are necessary, and these results might be interpreted within the context of the study limitations.

In the present study, neither ABCB1 nor ABCG2 allelic variants had a large influence on the AUC0–24 of O-desmethyl gefitinib in Japanese patients with NSCLC. P-glycoprotein and BCRP have been reported to contribute to the absorption and disposition of gefitinib [7, 16]; however, in a previous study and the present study [10, 14, 23], no associations between gefitinib exposure and ABCB1 or ABCG2 polymorphisms were found. Because O-desmethyl gefitinib has a higher solution than gefitinib, the transport of this metabolite may not be affected by ABC transporters. Consequently, these transporters do not seem to cause direct interindividual variability in O-desmethyl gefitinib pharmacokinetics. Our findings showed that the AUC0–24 of O-desmethyl gefitinib was significantly influenced only by CYP2D6 polymorphisms.

Conclusion

CYP2D6*5 and *10 polymorphisms are associated with the formation of O-desmethyl gefitinib from gefitinib. In CYP2D6 homozygous EMs, the plasma concentrations of O-desmethyl gefitinib were higher over 24 h after taking gefitinib than those of the parent compound; however, the side effects of diarrhoea, skin rash, and hepatotoxicity induced by gefitinib were unrelated to plasma exposure to O-desmethyl gefitinib.