Population Pharmacokinetics and Adverse Events of Erlotinib in Japanese Patients with Non-small-cell Lung Cancer: Impact of Genetic Polymorphisms in Metabolizing Enzymes and Transporters

Abstract

Determinants of interindividual variability in erlotinib pharmacokinetics (PK) and adverse events remain to be elucidated. This study with 50 Japanese non-small-cell lung cancer patients treated with oral erlotinib at a standard dose of 150 mg aimed to investigate whether genetic polymorphisms affect erlotinib PK and adverse events. Single nucleotide polymorphisms (SNPs) in genes encoding metabolizing enzymes (CYP1A1, CYP1A2, CYP2D6, CYP3A4, CYP3A5, UGT1A1, UGT2B7, GSTM1, and GSTT1) or efflux transporters (ABCB1, and ABCG2) were analyzed as covariates in a population PK model. The ABCB1 1236C>T (rs1128503) polymorphism, not ABCB1*2 haplotype (1236TT–2677TT–3455TT, rs1128503 TT–rs2032582 TT–rs1045642 TT), was a significant covariate for the apparent clearance (CL/F), with the TT genotype showing a 29.4% decrease in CL/F as compared with the CC and the CT genotypes. A marginally higher incidence of adverse events (mainly skin rash) was observed in the TT genotype group; however, patients with high plasma erlotinib exposure did not always experience skin rash. None of the other SNPs affected PK or adverse events. The ABCB1 genotype is a potential predictor for erlotinib adverse events. Erlotinib might be used with careful monitoring of adverse events in patients with ABCB1 polymorphic variants.

Erlotinib is an epidermal growth factor receptor (EGFR) tyrosine-kinase inhibitor, for which a daily dose of 150 mg has been shown to be effective for improving overall survival in non-small-cell lung cancer (NSCLC) patients who have failed at least 1 prior chemotherapy regimen.^1,2) Furthermore, erlotinib demonstrated significant prolongation of progression-free survival versus standard chemotherapy in EGFR mutation–positive NSCLC patients.^2,3) Based on these results, erlotinib monotherapy has received regulatory approval in over 100 countries, including countries in Europe, the United States, and Japan.

Erlotinib comes in tablets of 3 strengths (150, 100, 25 mg), and individual dose reduction following adverse events is recommended.^4,5) Skin rash and diarrhea are the most common adverse events, and these lead to discontinuation of therapy or dose reduction in many patients. It is reported in the U.S. package insert that skin rash occurred in 75% of patients receiving erlotinib as second- or third-line treatment for NSCLC,⁶⁾ and skin rash was observed in 98% of Japanese patients in clinical studies.^5,7,8) The incidence of skin rash is reported to be related to erlotinib exposure⁹⁾; therefore, a search for a dose regimen that has adequate efficacy while avoiding rash would greatly contribute to the improvement of a patient’s QOL. Diarrhea is a dose limiting toxicity and is probably related to drug-induced EGFR inhibition.¹⁰⁾ Hepatotoxicity is also a serious side effect, and precautions regarding hepatotoxicity are mentioned in the package insert in the United States and Japan.

Erlotinib is metabolized in the human liver, primarily by the CYP isoforms 3A4, 3A5, 1A1, and 1A2.^11–13) Erlotinib has been identified as a substrate of the ATP-binding cassette (ABC) transporters ABCB1 (P-glycoprotein/MDR1) and ABCG2 (BCRP).¹⁴⁾ The concentration of erlotinib in plasma has been shown to display large interindividual variability in clinical studies. Several population pharmacokinetics (PK) analyses have found that total bilirubin, alpha 1-acid glycoprotein, and smoking status affect its apparent clearance (CL/F).^9,15) Studies have also focused on polymorphisms of the CYP and ABC genes. Rudin et al.¹⁶⁾ and Phelps et al.¹⁷⁾ reported that polymorphisms of CYP3A4 and CYP3A5 tend to affect erlotinib CL/F. ABCG2 polymorphism has also been reported to influence erlotinib CL/F.¹⁸⁾ Our previous study showed that the ABCB1*2 haplotype (1236TT–2677TT–3455TT, rs1128503 genotype TT–rs2032582 genotype TT–rs1045642 genotype TT) impacted on erlotinib exposure in 28 Japanese patients.¹⁹⁾ However, these results are inconsistent and none of those analyses clearly explains the determinants of the large interindividual variability in the plasma concentration of erlotinib.

Here, with the aim of developing an erlotinib dose regimen that would minimize its adverse effects, we analyzed single nucleotide polymorphisms (SNPs) of the PK-related genes CYP1A1, CYP1A2, CYP3A4, CYP3A5, ABCB1, and ABCG2. We also analyzed CYP2D6 since erlotinib is partially metabolized by CYP2D6.²⁰⁾ UDP glucuronosyltransferase (UGT) 1A1 and UGT2B7 were analyzed since erlotinib inhibits these enzymes²¹⁾ thus erlotinib is possibly metabolized by these enzymes. Glutathione-S-transferase mu1 (GSTM1) and theta1 (GSTT1) genes which are reported to play important roles in hepatotoxicity^22,23) were analyzed since hepatotoxicity is a serious side effect of erlotinib. We selected all known SNPs before study initiation. We then examined the effect of these genetic polymorphisms on the PK of erlotinib by using population PK analysis. Furthermore, the relationships between these genotypes and adverse effects of erlotinib are reported here.

METHODS

Patients and Study Design

Fifty Japanese patients with NSCLC who were treated with oral erlotinib at a standard dose of 150 mg were enrolled in a prospective clinical study.¹⁹⁾ All patients had failed at least 1 prior chemotherapy regimen, and they were treated with erlotinib as a second-line or later treatment. Erlotinib was administered once daily in the morning and the dosage was not modified during the treatment period. Erlotinib treatment was continued until disease progression was observed, unacceptable adverse event became apparent, or the patient refused to continue receiving treatment. Co-medications which were known to affect erlotinib PK were prohibited during study treatment. This study was reviewed and approved by the relevant institutional review boards and signed informed consent was obtained from all patients prior to participation. These procedures were in accordance with the Helsinki Declaration.

Blood samples were collected from 28 patients at 5 to 18 time points (serial sampling at 0, 2, 4, 6, 8, 12, and 24 h after administration on day 1 and 8, and trough sampling at day 15, 29, and later), and trough blood samples at steady state (day 8 or later) were collected from 20 patients at a single time point.

Adverse events were assessed using the Common Terminology Criteria for Adverse Events (CTC AE version 3.0). Since adverse events might be confounded by the number of treatment periods, the grade of erlotinib-induced adverse event was determined during the first 30 days.

Genetic Polymorphism

Genomic DNA samples from 50 patients were obtained from whole blood by using a MagNA Pure LC DNA Isolation Kit I Tissue (Roche Diagnostics, Mannheim, Germany). Genotyping was then carried out using the following methods.

Applied Biosystems TaqMan SNP Genotyping Assay

Detailed explanations of the genotyping methods used in this study have been described in a previous report.²⁴⁾ The variant genotypes in Supplementary Material 1 were determined by using real-time quantitative PCR (qRT-PCR) on an ABI PRISM 7900HT system (Applied Biosystems, Foster City, CA, U.S.A.).

Direct Nucleotide Sequencing

The genotyping assays for the genes listed in Supplementary Material 2 were performed using the BigDye Terminator Cycle Sequencing Ready Reaction kit with an ABI PRISM 3130 Genetic Analyzer (Applied Biosystems). Primer sets for each gene are listed in Supplementary Material 2.^24,25)

PCR-Restriction Fragment Length Polymorphism (RFLP)

The gene alleles listed in Supplementary Material 3 were genotyped.^26–28) DNA fragments containing the polymorphic sites were amplified by PCR using the appropriate forward and reverse primers (Invitrogen LIFE Technologies Japan, Tokyo, Japan) (Supplementary Material 3). The products were then digested with restriction enzymes. The digested products were analyzed by electrophoresis on a 2% agarose gel in the presence of ethidium bromide. Genotype was determined based on individual fragment lengths.

Multiplex PCR

Deletions of GSTM1 and GSTT1 genes were analyzed by using multiplex PCR. The primers used are described in Supplementary Material 4.²⁹⁾ The PCR products were analyzed by electrophoresis on a 2% agarose gel in the presence of ethidium bromide. PCR products of 273, 480, and 493 base pairs indicated the presence of GSTM1, GSTT1, and β-actin, respectively.

Population Pharmacokinetics Model Development

Software

The population pharmacokinetics analysis of erlotinib was performed using a nonlinear mixed-effect model approach with NONMEM software (Version 7.2.0, ICON Development solutions, Dublin, Ireland.) All analyses were performed using the first-order conditional estimation method (FOCE) with interaction. R (version 3.1.1) was used for handling of data and results.

Pharmacokinetics Model

Blood samples were centrifuged and the plasma was frozen at −80°C until analysis. The concentrations of erlotinib in plasma were determined by using HPLC.³⁰⁾ Determination of the structural model was as described below. Model selection was based on changes in the objective function value (OFV; p<0.05). The difference in the OFV obtained by comparing each model is assumed to be asymptotically chi-squared distributed with degrees of freedom equal to the difference in the number of parameters between the 2 models.

Structural models

The following models were tested.

• Model 1: One-compartment model with first-order absorption and elimination.
• Model 2: Two-compartment model with first-order absorption and elimination.
• Model 3: Two-compartment model with first-order and lag-time absorption and elimination.
• Model 4: Two-compartment model with zero-order absorption and elimination.
• Model 5: Two-compartment model with first-order and zero-order absorption and elimination.

Error model

An exponential model Eq. 1 was evaluated to describe the interindividual variabilities of pharmacokinetic parameters:   

(1)

where θ_i represents the parameters of the ith subject, θ represents the population mean value of the parameter, and ηi is the random effect of the ith subject on the assumption of a normal distribution with a mean value of 0 and variance of ω².

A proportional error model Eq. 2 and a mixed error model Eq. 3 were tested as equations for the residual error:   

(2)

  

(3)

where C_ij represents the jth observed concentration of the ith subject, PRED_ij represents the jth predicted concentration of the ith subject, and ε_ij is the random effect of the jth observed concentration of the ith subject on the assumption of a normal distribution with a mean value of 0 and variance of δ².

Covariates of Pharmacokinetic Parameters

The influence of the covariates listed in Tables 1 and 2 was examined. Actual body weight before treatment initiation (listed in Table 1) was used as body weight. EGFR mutation status was excluded since no pharmacokinetic interaction was expected. Smoking status was excluded because most patients did not smoke during the study. In addition, ABCB1 1236C>T (rs1128503) was examined as C/C plus C/T vs. T/T. ABCB1 2677G>(T/A) (rs2032582) was examined as T/T vs. the others. The influence of continuous covariates was modeled according to Eq. 4, and that of categorical covariates was modeled according to Eq. 5:   

(4)

  

(5)

where PAR is the population mean value of the parameter, Cov is the value of the covariate, θ_Cov represents the slope of the effect of the covariate, PARCOV represents the individual estimate of the parameter, and Med_Cov represents the median value of the covariate in the study population. Investigation of the covariate effects on the CL/F was undertaken using stepwise covariate modeling.³¹⁾ The statistical level of significance was set to p<0.05 for forward inclusion and p<0.01 for backward deletion, and the final model was thereby determined.

Table 1. Patient Characteristics

Characteristic	n	%
Gender
Male	27	54
Female	23	46
Histological classification
Adenocarcinoma	41	82
Squamous cell carcinoma	7	14
Unknown	2	4
TNM classification
Stage 3	7	14
Stage 4	43	86
ECOG performance status
0	12	24
1	35	70
2	3	6
Smoking history
Never	23	46
Former	24	48
Current	3	6
Gefitinib experienced
No	29	58
Yes	21	42
EGFR mutation status
Wild type	8	16
Ex21 L858R	5	10
Ex19 deletion	7	14
Other	3	6
Unknown	27	54
Age (years)
Median [range]	68 [45–84]
Body weight (kg)
Median [range]	54 [35–75]
Body surface area (m2)
Median [range]	1.56 [1.20–1.86]
Total bilirubin (mg/dL)
Median [range]	0.55 [0.20–1.40]
Serum albumin (g/dL)
Median [range]	3.9 [1.3–4.9]
Serum creatinine (mg/dL)
Median [range]	0.73 [0.40–1.34]
Estimated glomerular filtration rate (mL/min/1.73 m2)
Median [range]	72.5 [39.5–114.8]

Table 2. Genotype Frequencies of Selected Genes Relevant to Erlotinib Disposition

Gene	rs Number	n	%	Gene	rs Number	n	%
ABCB1				CYP3A4
1236C>T	rs1128503			20230G>A	rs2242480
C/C		4	8	G/G		30	60
C/T		28	56	G/A		17	34
T/T		18	36	A/A		3	6
2677G>(T/A)	rs2032582			CYP3A5
G/A		6	12	6986A>G	rs776746
G/G		9	18	A/A		3	6
G/T		21	42	A/G		17	34
T/A		8	16	G/G		30	60
T/T		6	12	UGT1A1
3435C>T	rs1045642			*28	—
C/C		14	28	6/6		45	90
C/T		28	56	6/7		4	8
T/T		8	16	7/7		1	2
ABCG2				*6	rs4148323
−15622C>T	—			G/G		28	56
C/C		39	78	G/A		20	40
C/T		11	22	A/A		2	4
421C>A	rs2231142			UGT2B7
C/C		30	60	211G>T	rs12233719
C/A		17	34	G/G		39	78
A/A		3	6	G/T		7	14
1143C>T	rs2622604			T/T		4	8
C/C		38	76	802C>T	rs7439366
C/T		12	24	C/C		31	62
CYP1A1				C/T		14	28
2455A>G	rs1048943			T/T		5	10
A/A		32	64	GSTM1
A/G		12	24	Active		25	50
G/G		6	12	Null		25	50
CYP1A2				GSTT1
−3860G>A	rs2069514			Active		32	64
G/G		26	52	Null		18	36
G/A		20	40	ABCB1 haplotype
A/A		4	8	1236C>T–2677G>(T/A) −3435C>T: ABCB1*2
−2467T>delT	rs35694136			T/T-T/T-T/T (homo variant)		6	12
T/T		16	32	Other		44	88
T/delT		23	46	CYP1A2 haplotype
delT/delT		11	22	−3860G>A, −2467T>delT, 163C>A, 5347T>C
−163C>A	rs762551			A/A, delT/delT, A/A, C/C (homo variant)		4	8
C/C		4	8	Other		46	92
C/A		22	44	−2467T>delT, 5347T>C
A/A		24	48	delT/delT, C/C (homo variant)		11	22
5347T>C	rs2470890			Other		39	78
T/T		4	8
C/T		11	22
C/C		35	70
CYP2D6
100C>T	rs1065852
C/C		19	38
C/T		20	40
T/T		11	22

Model Evaluation

Goodness-of-fit plots were evaluated by visual inspection of diagnostic scatter plots. A nonparametric bootstrap resampling technique was adopted to validate the reliability and stability of the final model.¹⁵⁾ One thousand bootstrap replicates were generated by random resampling of the original data set with replacement. The stability and performance of the final model were assessed by using a visual predictive check (VPC). The VPC was performed 1000 times by simulating concentrations from the final model with the use of the original data set. The distribution of the observed concentrations was visually compared to the simulated distribution.

Statistical Analysis

The associations between genotypes and adverse event grades were determined by a Fisher’s exact test. An exploratory analysis was performed to assess time to appearance of grade 2 or higher skin rash by Kaplan–Meier methodology.

RESULTS

Population Characteristics and Genotype Frequencies of Selected Genes

PK data were available for a total of 48 of the 50 patients and pharmacogenetic data were available for all of the patients. Patient characteristics are summarized in Table 1. Approximately half of the patients were males (27 males vs. 23 females), and the median age was 68 years. The predominant histological classification was adenocarcinoma (>80%). The performance status of most patients (94%) was less than 2, and only 3 patients (6% of population) were current smokers (Table 1). The genotype frequencies of the selected genes are summarized in Table 2. All genotypes were in Hardy–Weinberg equilibrium, with the exception of ABCG2 −15622C>T, CYP1A1 2455 A>G (rs1048943), and UGT2B7 211G>T (rs12233719) (p<0.05).

Population PK Model

A two-compartment model with first-order absorption and elimination described the data well and was selected as the base model. Residual error was best characterized by a proportional error model. Parameter estimates are presented in Table 3. Twenty genotypes and 3 haplotypes, and 12 patient characteristics were individually tested by inclusion in the base model. Only ABCB1 1236C>T was a statically significant covariate for CL/F, with the TT genotype showing a 29.4% decrease in CL/F as compared with the CC and the CT genotypes (Table 3). The interindividual variability in CL/F decreased by 10.6% (from 41.5 to 37.1%) after inclusion of the TT genotype as a covariate in the model. In final model, the eta-shrinkage for CL/F, K_a, V₁/F, V₂/F and the epsilon-shrinkage were 11.3, 44.3, 29.8, 44.2, and 11.7%, respectively.

Table 3. Population Parameter Estimates

Parameter (Unit)	Final model	Bootstrap median (2.5th percentile, 97.5th percentile)	Base model
OFV	−217.854	—	−210.639
Typical CL/F (L·h−1)	3.64	3.62 (3.04, 4.29)	3.2
ABCB1 1236C>T on CL/F	−0.294	−0.299 (−0.45, −0.0963)	—
Typical Ka (h−1)	0.451	0.482 (0.0958, 0.961)	0.449
Typical V1/F (L)	67.2	68.1 (8.13, 103)	66.5
Typical V2/F (L)	41.1	41.5 (18.3, 74.7)	42
Typical Q (L·h−1)	8.97	7.86 (4.07, 21.6)	9.08
IIV CL/F (%)	37.1	36.3 (27.7, 43.5)	41.5
IIV Ka (%)	80.6	81.2 (21.0, 145)	79.4
IIV V1/F (%)	71.3	73.4 (43.7, 152)	71.4
IIV V2/F (%)	85.7	81.1 (37.8, 157)	83.8
REprop	0.0563	0.0548 (0.0385, 0.0730)	0.0563

OFV, objective function value; V₁, central volume of distribution; V₂, peripheral volume of distribution; Q, intercompartmental clearance; IIV CL/F, interindividual variability for CL/F; IIV V₁/F, interindividual variability for V₁/F; IIV V₂/F, interindividual variability for V₂/F; RE_prop, proportional component of residual error.

Goodness-of-fit plots obtained for the final model (Fig. 1) show that the final model fitting is acceptable and no systematic deviation over either time or population prediction value was observed. The medians and 2.5th and 97.5th percentiles for all parameter estimates obtained by the bootstrap method are listed in Table 3. The population PK parameter estimates in the bootstrap analyses were comparable to those generated using the original data set, indicating acceptable accuracy and stability of the structural and covariate models. The result of a visual-predictive check is shown in Fig. 2. Most of the observed concentrations fell within the 95% prediction interval. Overall, the final model was well able to describe the observed erlotinib concentrations.

Fig. 1. Goodness-of-Fit Plots for the Final Model

(a) Population predicted concentrations (PRED) vs. observed concentrations (DV). (b) Individual predicted concentrations (IPRED) vs. observed concentrations (DV). (c) Time after first administration (Time) vs. conditional weighted residuals (CWRES). (d) Population predicted concentrations (PRED) vs. conditional weighted residuals (CWRES).

Fig. 2. Visual Predictive Check

Open circles: observations; dashed lines: 97.5th percentile, median and 2.5th percentile of observed concentrations; solid lines: 97.5th percentile, median and 2.5th percentile of predicted concentrations.

Effect of SNPs on Adverse Events

None of the tested polymorphisms of the selected genes (CYP1A1, CYP1A2, CYP2D6, CYP3A4, CYP3A5, UGT1A1, UGT2B7, GSTM1, GSTT1, ABCB1, and ABCG2) showed significant association with adverse events. However, because ABCB1 1236C>T was the only statically significant covariate for CL/F of erlotinib, we examined the association of ABCB1 polymorphisms with the development of adverse events (Table 4). We analyzed the association between ABCB1 polymorphisms and cumulative incidence of adverse events (grade 3 or higher for all adverse events; grade 2 or higher for skin rash and diarrhea). We evaluated each genotype separately to explore the overall trend. The incidence of adverse events tended to be associated with the ABCB1 1236C>T polymorphism; however, the p-values of Fisher’s exact test were not significant (p=0.064 for all adverse events; p=0.189 for skin rash; p=0.590 for diarrhea). The p-values for the association of adverse events with ABCB1 2677G>(T/A) (p=0.167 for all adverse events; p=0.409 for skin rash; p=0.316 for diarrhea) or ABCB1 3435C>T (rs1045642) (p=0.228 for all adverse events; p=0.326 for skin rash; p=0.344 for diarrhea) were not significant.

Table 4. ABCB1 SNPs and Adverse Events

ABCB1 SNPs		n	All toxicity (%)							Skin rash (%)						Diarrhea (%)
			0	1	2	3	4	≥3 (%)	≥3 p-Value	0	1	2	3	≥2 (%)	≥2 p-Value	0	1	2	≥2 (%)	≥2 p-Value
1236C>T: rs1128503	C/C	4	1	2	1			0%	0.064	1	2	1		25%	0.189	4			0%	0.590
	C/T	28		10	16	1	1	7%		1	12	14	1	54%		25	1	2	7%
	T/T	18	2	3	7	4	2	33%		2	3	10	3	72%		11	4	3	17%
2677G>(T/A): rs2032582	G/A	6	1	2	2		1	17%	0.167	2	2	2		33%	0.409	6			0%	0.316
	G/G	9	1	2	5		1	11%		1	3	5		56%		7	1	1	11%
	G/T	21	1	9	8	3		14%		1	9	9	2	52%		16	3	2	10%
	T/A	8		2	6			0%			2	6		75%		8			0%
	T/T	6			3	2	1	50%			1	3	2	83%		3	1	2	33%
3435C>T: rs1045642	C/C	14	2	4	6		2	14%	0.228	3	5	6		43%	0.326	12	1	1	7%	0.344
	C/T	28	1	10	14	3		11%		1	10	15	2	61%		23	3	2	7%
	T/T	8		1	4	2	1	38%			2	4	2	75%		5	1	2	25%

The estimated median time to appearance of grade 2 or higher skin rash from date of study drug dosing was 10 d by Kaplan–Meier methodology. Time to first appearance of grade 2 or higher skin rash is shown in Supplementary Material 5. The median time to onset of grade 2 or higher skin rash appeared to be shorter in ABCB1 TT genotype groups for all ABCB1 SNPs than in the other genotype groups.

DISCUSSION

This study analyzed a greater number of genetic polymorphisms in candidate genes involved in both the PK pathway and erlotinib toxicity than has previously been attempted. The SNPs in CYP1A1 2455A>G, UGT2B7 211G>T (rs12233719), and ABCG2 (−15622C>T) deviated from Hardy–Weinberg equilibrium. However, the allele frequencies were similar to the frequencies reported in HapMap-JPT³²⁾ and in a previous report.¹⁶⁾

The population PK parameters obtained in this study were generally similar to those reported in other studies.^9,15–18) An exception was the PK parameters for total bilirubin, which, in contrast to previous reports,^9,15) was not a statistically significant covariate of CL/F. This discrepancy may be due to the different structural model or to the narrow range of total bilirubin values obtained in the present study.

ABCB1 1236C>T was the only statically significant covariate for CL/F of erlotinib, with the TT genotype showing a decrease in CL/F as compared with the CC and the CT genotypes. This is the first report to show the importance of ABCB1 on erlotinib PK by using population PK analysis. The 1236C>T polymorphism is suggested to minimize ABCB1 functionality in vitro.³³⁾ ABCB1 is highly expressed on the apical membrane of enterocytes in the entire intestine and on the bile canalicular membrane of hepatocytes, lower activity leads to higher absorption and lower elimination of erlotinib, thereby lowering CL/F. We previously reported that patients who were homozygous for the ABCB1*2 haplotype (1236TT–2677TT–3455TT) displayed higher plasma area under the curve (AUC) of erlotinib.¹⁹⁾ Interestingly, the 1236C>T SNP, but not the ABCB1*2 haplotype, had the strongest impact on, and was the only statistically significant covariate of, CL/F in the present study. In the previous report, data from only 28 patients were available and non-compartmental analysis was performed.¹⁹⁾ In the present study, we used data from 48 patients and we conducted population PK analysis. These differences between the two studies may have led to the different results obtained. The polymorphism data on ABCB1 is controversial and importance of ABCB1*2 haplotype over SNPs individually (1236C>T, 2677G>(T/A), and 3435C>T) is debatable. However, evidence is accumulating.^34–36) Both the ABCB1*2 and 1236C>T have significant effect on oxaliplatin³⁷⁾ susceptibility and sunitinib toxicities and survival.³⁸⁾ Megías-Vericat et al. reported the influence of each SNP on the effectiveness of cytarabine plus anthracyclines³⁹⁾ while He et al. emphasized on the impact of the haplotype on the effectiveness of the same therapy.⁴⁰⁾ The impact of the ABCB1 1236C>T SNP rather than the ABCB1*2 haplotype in this study is consistent with a previously published clinical study of methadone⁴¹⁾ and paclitaxel.⁴²⁾

ABCG2 421C>A (rs2231142) did not result in a significant decrease in CL/F, although a previous study has mentioned the importance of this SNP in a Japanese population.¹⁸⁾ It should be mentioned that the bioavailability of erlotinib increases in Bcrp1^−/−/Mdr1a/1b^−/− (ABCG2 and ABCB1 knockout) mice,¹⁴⁾ but that the PK of erlotinib is not substantially affected in ABCG2 knockout mice.⁴³⁾ Based on our results and the results of those mouse studies, ABCB1 could play a more important role than ABCG2 in erlotinib absorption.

ABCB1 polymorphisms were marginally associated with the incidence of adverse events. A marginally higher incidence of adverse events and earlier onset of grade 2 or higher skin rash were observed in the TT genotype group (Table 4, Supplementary Material 5). However, the incidence of adverse events in the TT genotype group of the ABCB1 1236C>T was not shown to be significantly different due to the lack of sufficient power for statistical analysis because patients were not enrolled based on their genotypes. In a previous report, a higher incidence of skin rash was considered to be related to an increase in the plasma exposure of erlotinib.⁹⁾ Although the present population PK analysis showed that the CL/F of erlotinib in the TT genotype group was lower than the CL/F of erlotinib in the other groups, patients with high plasma erlotinib exposure did not always experience skin rash. Therefore, the erlotinib AUC might not directly cause the skin rash.

ABCB1 is expressed in sweat ducts, blood vessels, nerve sheaths, and muscles of human skin.⁴⁴⁾ The tissue-to-plasma concentration ratio in the skin of mdr1a/1b^−/− mice is reported to be significantly higher than that in wild-type mice.⁴⁵⁾ Since the TT genotype of ABCB1 has less activity, the erlotinib concentration in the skin of the TT genotype patients is presumably much higher than that in patients of the other genotypes, and thereby causes skin rash. Furthermore, the higher frequency of TT genotype in Asian population can reasonably explain higher incidence of skin rash in Japanese compared with the other populations,^5–8) whereas the incidence of grade 2 or higher skin rash were not reported to differ significantly between Asians and non-Asians.⁴⁶⁾ Our result is also consistent with the recent report which shows that ABCB1 1236C>T TT genotype had greater risk of skin rash after the treatment of other EGFR tyrosine-kinase inhibitor, gefitinib.⁴⁷⁾ Therefore, polymorphism analyses of ABCB1 1236C>T might be beneficial to predict the incidence of skin rash in erlotinib treatment. Nevertheless, erlotinib can be safely used for TT genotype patients provided there is careful monitoring of adverse events.

In this study, 1 patient developed grade 4 hepatotoxicity. This hepatotoxicity was not expected to be related to either the plasma exposure of erlotinib or to GST genotypes. The plasma ophthalmic acid concentrations of 26 patients at baseline (before erlotinib treatment) were determined by liquid chromatography tandem mass spectrometry as described in a previous report.⁴⁸⁾ Interestingly, the patient who developed grade 4 hepatotoxicity had a high level of ophthalmic acid (36.9 ng/mL), while the other 25 patients, who did not develop hepatotoxicity, had no detectable (<2 ng/mL) ophthalmic acid. Ophthalmic acid is an analog of glutathione, which acts as an enzyme-catalyzed antioxidant when cells and organisms experience electrophilic stress that arises from metabolic processes.⁴⁹⁾ This acid is a sensitive indicator of glutathione levels in a mouse model, with a high level of this acid indicating glutathione depletion.⁵⁰⁾ Furthermore, ophthalmic acid in the serum of paracetamol-induced acute liver failure patients has been reported to be detected more frequently in non-survivors.⁵¹⁾ Therefore, the above-described case in our study indicates the possibility that a patient with higher oxidative stress prior to treatment might have a higher risk of developing severe erlotinib-induced hepatotoxicity.

The present study has several limitations. The sample size of this study is smaller than that of a previous report¹⁸⁾ (n=50 vs. n=88), and the study was of an observational case-control nature, therefore effect of SNPs on adverse events could not been analyzed by Cox proportional hazard model. The effect of smoking status was not tested because all former smokers had quit smoking more than 2 weeks before the initiation of study treatment and therefore the effect of smoking on inducing CYP1A2 activity was negligible.⁵²⁾ The EGFR polymorphisms which were reported to be associated with erlotinib or gefitinib toxicities^16,53) were not evaluated, since the EGFR mutation test was not mandatory for second-line or later treatment of erlotinib. However, our study employed concentration data from more than 50 plasma samples collected after day 14 of the first erlotinib administration, and we were able to estimate CL/F of erlotinib over time, which was not clearly estimated in the previous report.¹⁸⁾ Further, randomized controlled trials with genotyped patients are needed to clarify the relationship between ABCB1 genotype, tissue erlotinib concentration, and adverse effects.

In conclusion, only ABCB1 1236C>T influenced exposure to erlotinib. This SNP was suggested to be related to the risk of adverse events. Erlotinib might be used with careful monitoring of adverse events for patients with ABCB1 polymorphic variants. The effects of SNPs in other genes (CYP1A1, CYP1A2, CYP2D6, CYP3A4, CYP3A5, UGT1A1, UGT2B7, GSTM1, GSTT1, and ABCG2) were denied. Further clinical studies are needed to confirm the ABCB1 impact and to investigate the mechanism underlying how ABCB1 affects the incidence of adverse events.

Acknowledgment

The authors thank Satofumi Iida (Chugai Pharmaceutical Co., Ltd.) for his assistance with the population PK analysis.

Conflict of Interest

This study was supported in part by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (AH: number, KAKENHI 21590167). C.E. is an employee of Chugai Pharmaceutical Co., Ltd. A.H. has received research funding from Chugai Pharmaceutical Co., Ltd. J.S., S.S., N.I., M.I., S.U., H. Kishi, S.F., H. Semba, K.K., Y.T., M.H., Y.K, H. Saito, T.I., and H. Kohrogi declare no conflict of interest.

Supplementary Materials

The online version of this article contains supplementary materials.

REFERENCES

• 1) Shepherd FA, Rodrigues Pereira J, Ciuleanu T, Tan EH, Hirsh V, Thongprasert S, Campos D, Maoleekoonpiroj S, Smylie M, Martins R, van Kooten M, Dediu M, Findlay B, Tu D, Johnston D, Bezjak A, Clark G, Santabarbara P, Seymour L. Erlotinib in previously treated non-small-cell lung cancer. N. Engl. J. Med., 353, 123–132 (2005).
• 2) Ellis PM, Coakley N, Feld R, Kuruvilla S, Ung YC. Use of the epidermal growth factor receptor inhibitors gefitinib, erlotinib, afatinib, dacomitinib, and icotinib in the treatment of non-small-cell lung cancer: a systematic review. Curr. Oncol., 22, e183–e215 (2015).
• 3) Rosell R, Carcereny E, Gervais R, Vergnenegre A, Massuti B, Felip E, Palmero R, Garcia-Gomez R, Pallares C, Sanchez JM, Porta R, Cobo M, Garrido P, Longo F, Moran T, Insa A, De Marinis F, Corre R, Bover I, Illiano A, Dansin E, de Castro J, Milella M, Reguart N, Altavilla G, Jimenez U, Provencio M, Moreno MA, Terrasa J, Munoz-Langa J, Valdivia J, Isla D, Domine M, Molinier O, Mazieres J, Baize N, Garcia-Campelo R, Robinet G, Rodriguez-Abreu D, Lopez-Vivanco G, Gebbia V, Ferrera-Delgado L, Bombaron P, Bernabe R, Bearz A, Artal A, Cortesi E, Rolfo C, Sanchez-Ronco M, Drozdowskyj A, Queralt C, de Aguirre I, Ramirez JL, Sanchez JJ, Molina MA, Taron M, Paz-Ares L. Erlotinib versus standard chemotherapy as first-line treatment for European patients with advanced EGFR mutation-positive non-small-cell lung cancer (EURTAC): a multicentre, open-label, randomised phase 3 trial. Lancet Oncol., 13, 239–246 (2012).
• 4) Binder D, Buckendahl AC, Hubner RH, Schlattmann P, Temmesfeld-Wollbruck B, Beinert T, Suttorp N. Erlotinib in patients with advanced non-small-cell lung cancer: impact of dose reductions and a novel surrogate marker. Med. Oncol., 29, 193–198 (2012).
• 5) Reck M, Mok T, Wolf J, Heigener D, Wu YL. Reviewing the safety of erlotinib in non-small cell lung cancer. Expert Opin. Drug Saf., 10, 147–157 (2011).
• 6) “Genentech USA Tarceva (erlotinib) tablets: US prescribing information.”: ‹https://www.gene.com/download/pdf/tarceva_prescribing.pdf ›.
• 7) Kubota K, Nishiwaki Y, Tamura T, Nakagawa K, Matsui K, Watanabe K, Hida T, Kawahara M, Katakami N, Takeda K, Yokoyama A, Noda K, Fukuoka M, Saijo N. Efficacy and safety of erlotinib monotherapy for Japanese patients with advanced non-small cell lung cancer: a phase II study. J. Thorac. Oncol., 3, 1439–1445 (2008).
• 8) Takahashi T, Yamamoto N, Nukiwa T, Mori K, Tsuboi M, Horai T, Masuda N, Eguchi K, Mitsudomi T, Yokota S, Segawa Y, Ichinose Y, Fukuoka M, Saijo N. Phase II study of erlotinib in Japanese patients with advanced non-small cell lung cancer. Anticancer Res., 30, 557–563 (2010).
• 9) Lu JF, Eppler SM, Wolf J, Hamilton M, Rakhit A, Bruno R, Lum BL. Clinical pharmacokinetics of erlotinib in patients with solid tumors and exposure-safety relationship in patients with non-small cell lung cancer. Clin. Pharmacol. Ther., 80, 136–145 (2006).
• 10) Hidalgo M, Siu LL, Nemunaitis J, Rizzo J, Hammond LA, Takimoto C, Eckhardt SG, Tolcher A, Britten CD, Denis L, Ferrante K, Von Hoff DD, Silberman S, Rowinsky EK. Phase I and pharmacologic study of OSI-774, an epidermal growth factor receptor tyrosine kinase inhibitor, in patients with advanced solid malignancies. J. Clin. Oncol., 19, 3267–3279 (2001).
• 11) Ling J, Johnson KA, Miao Z, Rakhit A, Pantze MP, Hamilton M, Lum BL, Prakash C. Metabolism and excretion of erlotinib, a small molecule inhibitor of epidermal growth factor receptor tyrosine kinase, in healthy male volunteers. Drug Metab. Dispos., 34, 420–426 (2006).
• 12) Rakhit A, Pantze MP, Fettner S, Jones HM, Charoin JE, Riek M, Lum BL, Hamilton M. The effects of CYP3A4 inhibition on erlotinib pharmacokinetics: computer-based simulation (SimCYP) predicts in vivo metabolic inhibition. Eur. J. Clin. Pharmacol., 64, 31–41 (2008).
• 13) Hamilton M, Wolf JL, Drolet DW, Fettner SH, Rakhit AK, Witt K, Lum BL. The effect of rifampicin, a prototypical CYP3A4 inducer, on erlotinib pharmacokinetics in healthy subjects. Cancer Chemother. Pharmacol., 73, 613–621 (2014).
• 14) Marchetti S, de Vries NA, Buckle T, Bolijn MJ, van Eijndhoven MA, Beijnen JH, Mazzanti R, van Tellingen O, Schellens JH. Effect of the ATP-binding cassette drug transporters ABCB1, ABCG2, and ABCC2 on erlotinib hydrochloride (Tarceva) disposition in in vitro and in vivo pharmacokinetic studies employing Bcrp1−/−/Mdr1a/1b−/− (triple-knockout) and wild-type mice. Mol. Cancer Ther., 7, 2280–2287 (2008).
• 15) Emoto-Yamamoto Y, Iida S, Kawanishi T, Fukuoka M. Population pharmacokinetics of erlotinib in Japanese patients with advanced non-small cell lung cancer. J. Clin. Pharm. Ther., 40, 232–239 (2015).
• 16) Rudin CM, Liu W, Desai A, Karrison T, Jiang X, Janisch L, Das S, Ramirez J, Poonkuzhali B, Schuetz E, Fackenthal DL, Chen P, Armstrong DK, Brahmer JR, Fleming GF, Vokes EE, Carducci MA, Ratain MJ. Pharmacogenomic and pharmacokinetic determinants of erlotinib toxicity. J. Clin. Oncol., 26, 1119–1127 (2008).
• 17) Phelps MA, Stinchcombe TE, Blachly JS, Zhao W, Schaaf LJ, Starrett SL, Wei L, Poi M, Wang D, Papp A, Aimiuwu J, Gao Y, Li J, Otterson GA, Hicks WJ, Socinski MA, Villalona-Calero MA. Erlotinib in African Americans with advanced non-small cell lung cancer: a prospective randomized study with genetic and pharmacokinetic analyses. Clin. Pharmacol. Ther., 96, 182–191 (2014).
• 18) Fukudo M, Ikemi Y, Togashi Y, Masago K, Kim YH, Mio T, Terada T, Teramukai S, Mishima M, Inui K, Katsura T. Population pharmacokinetics/pharmacodynamics of erlotinib and pharmacogenomic analysis of plasma and cerebrospinal fluid drug concentrations in Japanese patients with non-small cell lung cancer. Clin. Pharmacokinet., 52, 593–609 (2013).
• 19) Hamada A, Sasaki J, Saeki S, Iwamoto N, Inaba M, Ushijima S, Urata M, Kishi H, Fujii S, Semba H, Kashiwabara K, Tsubata Y, Kai Y, Isobe T, Kohrogi H, Saito H. Association of ABCB1 polymorphisms with erlotinib pharmacokinetics and toxicity in Japanese patients with non-small-cell lung cancer. Pharmacogenomics, 13, 615–624 (2012).
• 20) Li J, Zhao M, He P, Hidalgo M, Baker SD. Differential metabolism of gefitinib and erlotinib by human cytochrome P450 enzymes. Clin. Cancer Res., 13, 3731–3737 (2007).
• 21) Liu Y, Ramirez J, House L, Ratain MJ. Comparison of the drug-drug interactions potential of erlotinib and gefitinib via inhibition of UDP-glucuronosyltransferases. Drug Metab. Dispos., 38, 32–39 (2010).
• 22) Yiannakopoulou EC. Pharmacogenomics of phase II metabolizing enzymes and drug transporters: clinical implications. Pharmacogenomics J., 13, 105–109 (2013).
• 23) Kurose K, Sugiyama E, Saito Y. Population differences in major functional polymorphisms of pharmacokinetics/pharmacodynamics-related genes in Eastern Asians and Europeans: implications in the clinical trials for novel drug development. Drug Metab. Pharmacokinet., 27, 9–54 (2012).
• 24) Yamakawa Y, Hamada A, Nakashima R, Yuki M, Hirayama C, Kawaguchi T, Saito H. Association of genetic polymorphisms in the influx transporter SLCO1B3 and the efflux transporter ABCB1 with imatinib pharmacokinetics in patients with chronic myeloid leukemia. Ther. Drug Monit., 33, 244–250 (2011).
• 25) Fujita K, Ando Y, Yamamoto W, Miya T, Endo H, Sunakawa Y, Araki K, Kodama K, Nagashima F, Ichikawa W, Narabayashi M, Akiyama Y, Kawara K, Shiomi M, Ogata H, Iwasa H, Okazaki Y, Hirose T, Sasaki Y. Association of UGT2B7 and ABCB1 genotypes with morphine-induced adverse drug reactions in Japanese patients with cancer. Cancer Chemother. Pharmacol., 65, 251–258 (2010).
• 26) He BX, Shi L, Qiu J, Tao L, Li R, Yang L, Zhao SJ. A functional polymorphism in the CYP3A4 gene is associated with increased risk of coronary heart disease in the Chinese Han population. Basic Clin. Pharmacol. Toxicol., 108, 208–213 (2011).
• 27) Chen X, Wang L, Zhi L, Zhou G, Wang H, Zhang X, Hao B, Zhu Y, Cheng Z, He F. The G-113A polymorphism in CYP1A2 affects the caffeine metabolic ratio in a Chinese population. Clin. Pharmacol. Ther., 78, 249–259 (2005).
• 28) Pavanello S, Pulliero A, Lupi S, Gregorio P, Clonfero E. Influence of the genetic polymorphism in the 5′-noncoding region of the CYP1A2 gene on CYP1A2 phenotype and urinary mutagenicity in smokers. Mutat. Res., 587, 59–66 (2005).
• 29) García-Closas M, Kelsey KT, Hankinson SE, Spiegelman D, Springer K, Willett WC, Speizer FE, Hunter DJ. Glutathione S-transferase mu and theta polymorphisms and breast cancer susceptibility. J. Natl. Cancer Inst., 91, 1960–1964 (1999).
• 30) Lepper ER, Swain SM, Tan AR, Figg WD, Sparreboom A. Liquid-chromatographic determination of erlotinib (OSI-774), an epidermal growth factor receptor tyrosine kinase inhibitor. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci., 796, 181–188 (2003).
• 31) Jonsson EN, Karlsson MO. Automated covariate model building within NONMEM. Pharm. Res., 15, 1463–1468 (1998).
• 32) “The National Center for Biotechnology Information (NCBI).” ‹https://www.ncbi.nlm.nih.gov/projects/SNP/ ›.
• 33) Salama NN, Yang Z, Bui T, Ho RJ. MDR1 haplotypes significantly minimize intracellular uptake and transcellular P-gp substrate transport in recombinant LLC-PK1 cells. J. Pharm. Sci., 95, 2293–2308 (2006).
• 34) McLean C, Wilson A, Kim RB. Impact of Transporter Polymorphisms on Drug Development: Is It Clinically Significant? J. Clin. Pharmacol., 56 (Suppl. 7), S40–S58 (2016).
• 35) Tulsyan S, Mittal RD, Mittal B. The effect of ABCB1 polymorphisms on the outcome of breast cancer treatment. Pharmacogenomics and Personalized Medicine, 9, 47–58 (2016).
• 36) Onodera M, Endo K, Kakuta Y, Kuroha M, Kimura T, Hiramoto K, Kanazawa Y, Negoro K, Shiga H, Kinouchi Y, Shimosegawa T. ATP-binding cassette subfamily B member 1 1236C/T polymorphism significantly affects the therapeutic outcome of tacrolimus in patients with refractory ulcerative colitis. J. Gastroenterol. Hepatol., 32, 1562–1569 (2017).
• 37) Wu H, Kang H, Liu Y, Xiao Q, Zhang Y, Sun M, Liu D, Wang Z, Zhao H, Yao W, Jia T, Wang E, Zheng Z, Wei M. Association of ABCB1 genetic polymorphisms with susceptibility to colorectal cancer and therapeutic prognosis. Pharmacogenomics, 14, 897–911 (2013).
• 38) Chu YH, Li H, Tan HS, Koh V, Lai J, Phyo WM, Choudhury Y, Kanesvaran R, Chau NM, Toh CK, Ng QS, Tan PH, Chowbay B, Tan MH. Association of ABCB1 and FLT3 Polymorphisms with Toxicities and Survival in Asian Patients Receiving Sunitinib for Renal Cell Carcinoma. PLOS ONE, 10, e0134102 (2015).
• 39) Megías-Vericat JE, Rojas L, Herrero MJ, Boso V, Montesinos P, Moscardo F, Poveda JL, Sanz MA, Alino SF. Influence of ABCB1 polymorphisms upon the effectiveness of standard treatment for acute myeloid leukemia: a systematic review and meta-analysis of observational studies. Pharmacogenomics J., 15, 109–118 (2015).
• 40) He H, Yin J, Li X, Zhang Y, Xu X, Zhai M, Chen J, Qian C, Zhou H, Liu Z. Association of ABCB1 polymorphisms with prognostic outcomes of anthracycline and cytarabine in Chinese patients with acute myeloid leukemia. Eur. J. Clin. Pharmacol., 71, 293–302 (2015).
• 41) Levran O, O’Hara K, Peles E, Li D, Barral S, Ray B, Borg L, Ott J, Adelson M, Kreek MJ. ABCB1 (MDR1) genetic variants are associated with methadone doses required for effective treatment of heroin dependence. Hum. Mol. Genet., 17, 2219–2227 (2008).
• 42) Shimizu C, Tanabe Y, Hamada A, Yunokawa M, Takano T, Koganemaru S, Ozaki Y, Tamura N, Fujiwara Y, Tamura K. Association of paclitaxel-induced sensory peripheral neuropathy with the ABCB1 genetic variant and age. J. Clin. Oncol., 33, 11100 (2015).
• 43) Elmeliegy MA, Carcaboso AM, Tagen M, Bai F, Stewart CF. Role of ATP-binding cassette and solute carrier transporters in erlotinib CNS penetration and intracellular accumulation. Clin. Cancer Res., 17, 89–99 (2011).
• 44) Skazik C, Wenzel J, Marquardt Y, Kim A, Merk HF, Bickers DR, Baron JM. P-glycoprotein (ABCB1) expression in human skin is mainly restricted to dermal components. Exp. Dermatol., 20, 450–452 (2011).
• 45) Ito K, Nguyen HT, Kato Y, Wakayama T, Kubo Y, Iseki S, Tsuji A. P-glycoprotein (Abcb1) is involved in absorptive drug transport in skin. J. Control. Release, 131, 198–204 (2008).
• 46) Takeda M, Okamoto I, Nakagawa K. Pooled safety analysis of EGFR-TKI treatment for EGFR mutation-positive non-small cell lung cancer. Lung Cancer, 88, 74–79 (2015).
• 47) Ma Y, Xin S, Huang M, Yang Y, Zhu C, Zhao H, Zhang Y, Chen L, Zhao Y, Li J, Zhuang W, Zhu X, Zhang L, Wang X. Determinants of Gefitinib toxicity in advanced non-small cell lung cancer (NSCLC): a pharmacogenomic study of metabolic enzymes and transporters. Pharmacogenomics J., 17, 325–330 (2017).
• 48) Geenen S, Michopoulos F, Kenna JG, Kolaja KL, Westerhoff HV, Wilson I. HPLC-MS/MS methods for the quantitative analysis of ophthalmic acid in rodent plasma and hepatic cell line culture medium. J. Pharm. Biomed. Anal., 54, 1128–1135 (2011).
• 49) Dello SA, Neis EP, de Jong MC, van Eijk HM, Kicken CH, Olde Damink SW, Dejong CH. Systematic review of ophthalmate as a novel biomarker of hepatic glutathione depletion. Clin. Nutr., 32, 325–330 (2013).
• 50) Soga T, Baran R, Suematsu M, Ueno Y, Ikeda S, Sakurakawa T, Kakazu Y, Ishikawa T, Robert M, Nishioka T, Tomita M. Differential metabolomics reveals ophthalmic acid as an oxidative stress biomarker indicating hepatic glutathione consumption. J. Biol. Chem., 281, 16768–16776 (2006).
• 51) Kaur G, Leslie EM, Tillman H, Lee WM, Swanlund DP, Karvellas CJ. Detection of Ophthalmic Acid in Serum from Acetaminophen-Induced Acute Liver Failure Patients Is More Frequent in Non-Survivors. PLOS ONE, 10, e0139299 (2015).
• 52) Faber MS, Fuhr U. Time response of cytochrome P450 1A2 activity on cessation of heavy smoking. Clin. Pharmacol. Ther., 76, 178–184 (2004).
• 53) Giovannetti E, Zucali PA, Peters GJ, Cortesi F, D’Incecco A, Smit EF, Falcone A, Burgers JA, Santoro A, Danesi R, Giaccone G, Tibaldi C. Association of polymorphisms in AKT1 and EGFR with clinical outcome and toxicity in non-small cell lung cancer patients treated with gefitinib. Mol. Cancer Ther., 9, 581–593 (2010).