PXR Pharmacogenetics: Association of Haplotypes with Hepatic CYP3A4 and ABCB1 Messenger RNA Expression and Doxorubicin Clearance in Asian Breast Cancer Patients

Purpose: To characterize pregnane X receptor (PXR) polymorphic variants in healthy Asian populations [Chinese, Malay and Indian (n = 100 each)], and to investigate the association between PXR haplotypes and hepatic mRNA expression of PXR and its downstream target genes, CYP3A4 and ABCB1, as well as their influence on the clearance of doxorubicin in Asian breast cancer patients.

Experimental Design:PXR genotyping was done by direct DNA sequencing, and PXR haplotypes and haplotype clusters were derived by expectation-maximization algorithm. Genotype-phenotype correlations were done using Mann-Whitney U test and Kruskal-Wallis test.

Results: Significant interethnic variations were observed in PXR pharmacogenetics among the three Asian ethnic groups. The expression of PXR mRNA in liver tissues harboring the PXR*1B haplotype clusters was 4-fold lower compared with the non-PXR*1B (*1A + *1C) haplotype clusters [PXR*1B versus PXR*1A; P = 0.015; PXR*1B versus PXR*1C; P = 0.023]. PXR*1B-bearing liver tissues were associated with significantly lower expression of CYP3A4 (PXR*1B versus non-PXR*1B, P = 0.030) and ABCB1 (PXR*1B versus non-PXR*1B, P = 0.060) compared with non–PXR*1B-bearing liver tissues. Doxorubicin clearance in breast cancer patients harboring the PXR*1B haplotypes was significantly lower compared with patients carrying the non-PXR*1B haplotypes [PXR*1B versus non-PXR*1B, CL/BSA (L h⁻¹ m⁻²): 20.84 (range, 8.68-29.24) versus 24.85 (range, 13.80-55.66), P = 0.022].

Conclusions: This study showed that PXR*1B was associated with reduced hepatic mRNA expression of PXR and its downstream targets, CYP3A4 and ABCB1. Genotype-phenotype correlates in breast cancer patients showed PXR*1B to be significantly associated with lower doxorubicin clearance, suggesting that PXR haplotype constitution could be important in influencing interindividual and interethnic variations in disposition of its putative drug substrates.

Nuclear receptors act as sensors for endogenous and exogenous compounds by transferring stimuli into cellular responses and regulating the expression of their target genes. The nuclear receptors comprise a family of ligand-activated transcription factors that include the steroid, retinoid, and thyroid hormone receptors and have critical roles in nearly every aspect of developmental and adult physiology (1). The discovery of the orphan nuclear receptor pregnane X receptor (PXR; NR1I2), also referred to as steroid and xenobiotic receptor, provided insights into the molecular basis of how specific drugs could induce pathways of clearance and enhance rate of elimination of drugs (2).

The PXR gene is ∼40 kb and maps to chromosome 3 (3q12-q13.3). The human PXR (hPXR) mRNA consists of nine exons and encodes a protein with a molecular weight of 49.7 kDa and consists of 434 amino acids. Ten alternatively spliced forms of hPXR have been identified to date (2, 3). hPXR.1 contains an activation function-1 region (amino acid residues 1-40), a DNA binding domain (amino acid residues 41-140), and a ligand binding domain (amino acid residues 141-434). The hPXR.2 cDNA encodes a 390-amino-acid protein that lacks a 41-amino-acid region in the putative ligand binding domain of hPXR.1. The hPXR mRNA is predominantly expressed in liver and small intestine and, to a lesser extent, in kidneys, lungs, and breast (4).

A wide variety of structurally divergent endogenous compounds and therapeutic agents have been identified as ligands of PXR, which include steroid hormones and anti-inflammatory, antifungal, antihypertensive, antiviral, and antineoplastic agents (2, 5). In silico screening have identified putative PXR/RXR binding sites in the regulatory regions of ∼281 genes involved in encoding drug-metabolizing enzymes and 97 genes involved in encoding drug transporters (6). The activation of PXR is initiated in the cytoplasm, and following exposure to ligands, heterodimers consisting of PXR and the retinoic acid receptor (RXR; NRIB2) bind to nuclear receptor response elements in the upstream regulatory regions of target genes encoding drug-metabolizing enzymes such as CYP3A4 and drug transporters such as the ATP binding cassette (ABC) superfamily of transporters to induce their transcription (7–9).

Polymorphisms in the promoter, DNA binding domain, and ligand binding domain regions of the PXR gene have been reported to have significant functional consequences, including aberrant DNA binding and alterations in transactivation and expression of downstream target genes (10–13). The PXR −1570C>T polymorphism lies within a nuclear factor-κB binding element in the promoter region and the −1570TT genotype has previously been shown to be associated with higher rifampicin induction of CYP3A4 whereas the 5′-untranslated region (UTR) −566A>C polymorphism has been related to increased ABCB1 and CYP3A4 mRNA expression (10, 11). The exon 4 PXR 365G>A (R122Q) and 488A>G (D163G) polymorphisms have also been associated with attenuation of rifampicin-induced CYP3A4 promoter activity (10, 12).

Identifying the role of functional polymorphisms in the PXR gene and their prevalence in different populations may thus help explain the mechanistic basis of variations in the pharmacokinetics and pharmacodynamics of drugs that are PXR ligands. Such PXR-related genotypic-phenotypic correlative studies are lacking in the Asian population and would be useful in further enhancing the understanding of the mechanistic basis of interindividual and interethnic variability in drug metabolism and disposition (3). The objectives of this study were thus twofold: (a) to study the pharmacogenetics of PXR gene in healthy subjects belonging to three distinct Asian ethnic groups (Chinese, Malays, and Indians); (b) to investigate the association between PXR haplotypes and hepatic expression of PXR and its downstream target genes, CYP3A4 and ABCB1, as well as their influence on the clearance of doxorubicin, an ABCB1 substrate in Asian breast cancer patients.

Subjects. The healthy subjects recruited for the study comprised three ethnic groups predominant in the Asian population [Chinese, Malays, and Indians (n = 100 in each group)]. The ethnicity of the study subjects was confirmed by careful screening and verified against their National Registry Identification Cards. The patients recruited for the study had histologically confirmed invasive breast cancer (n = 62) and deemed fit to receive adjuvant chemotherapy with doxorubicin (A)/cyclophosphamide(C). All participants in the study provided approved informed consent for the study, which was approved by the ethics review committee of the National Cancer Center, Singapore.

Pharmacokinetics of doxorubicin in breast cancer patients. Doxorubicin was administered at a dose of 60 mg/m² i.v. over 20 min and cyclophosphamide at 600 mg/m² i.v. over 30 min at once every 3-wk cycle. Standard premedications were uniformly prescribed to all patients and included dexamethasone (10 mg), ranitidine (50 mg), and diphenhydramine (50 mg). The patients did not receive any other anticancer cytotoxic therapy, immunotherapy, or biological response modifiers during the study. Blood samples for pharmacokinetic analysis were drawn from breast cancer patients at the following time points on the first day of the first cycle of chemotherapy: at predose; 5, 15, and 30 min; and 1, 4, 8, and 24 h. Plasma concentrations of doxorubicin were estimated by reverse-phase high-performance liquid chromatography with fluorescence detection as described in earlier reports (14, 15). Briefly, following a single protein precipitation step, chromatographic separation was accomplished using a C-18 column with a mobile phase consisting of 50 mmol/L sodium phosphate buffer/acetonitrile/1-propanol (65:25:2, v/v/v), pH 2.0. The analytes were measured by fluorescence detection with excitation wavelength of 480 nm and emission wavelength of 560 nm. The pharmacokinetic parameters of doxorubicin were determined using the nonlinear regression program WinNonLin version 2.1 (Pharsight, Inc.).

Pharmacogenetic analysis. Purified genomic DNA was extracted from peripheral leucocytes (5 mL) of healthy subjects and liver tissues by using the DNeasy Blood and Tissue kit (Qiagen, Inc.). Genotyping involved screening of polymorphisms in nine exons of the PXR gene (GenBank accession no. AF364606) including the exon-intron junctions and 2 kb of the 5′- and 3′-UTR regions, respectively (primer sequences are available in Supplementary Table S1). The sequencing reactions and base calling of resultant trace files were done using Applied Biosystems 3730 DNA Analyzer (Applied Biosystems).

Linkage disequilibrium and haplotype analysis. Pairwise linkage disequilibrium (LD) between the polymorphisms was quantified by |D′| and rho square (r²) values, and LD blocks were identified within the PXR region (Haploview software, version 3.32, Daly lab, Broad Institute; ref. 16). |D′| is the coefficient of LD, which measures the nonrandom association of polymorphisms at two or more loci, and r² refers to correlation coefficient between two alleles. The haplotype structure of each ethnic group was inferred by maximum likelihood estimation based on the expectation-maximization algorithm. Network analysis was done on haplotypes with a frequency of >0.01 to further clarify the relationships among haplotypes using Network software version 4.2 (17).⁵

Median-joining algorithm was implemented to calculate the network that supports multistate data (A, C, G, and T) to reconstruct all possible shortest least complex trees, thereby making parental nodes and their neighboring haplotypes (haplotype cluster). Haplotype associations through network-based analysis hold the assumption that constituent haplotypes have minimal intravariation and large intervariations due to their sequence similarities (r² > 0.75; ref. 17).

Hepatic expression. Healthy, noncancerous liver tissues (n = 41) were available from Chinese cancer patients undergoing hepatectomy for metastasis from a colonic primary malignancy and were certified to be free of malignant cells by pathologic examinations. Total RNA was extracted from the liver tissues using TRIzol reagent (Invitrogen Corp.), and reverse transcription was conducted in a reaction mix containing 10 μL of the extracted total RNA (10 μg/mL), 1× Promega avian myeloblastosis virus reverse transcriptase reaction, 1 mmol/L of deoxynucleotide triphosphates, 0.5 μg of random hexamer, 40 units of RNase inhibitor, and 4.5 μL of Promega avian myeloblastosis virus reverse transcriptase (Promega Corp.).

The forward and reverse primers used for reverse transcription-PCR analysis were PXR (forward: 5′-GGACGCTCAGATGAAAACCTTT-3′ and reverse: 5′-GCAGC CGGAAATTCTTGAAA-3′), CYP3A4 (forward: 5′-TCATTGCTGTCT CCAACCTTCA-3′ and reverse: 5′-GCTTCCCGCCTCAGATTTCT-3′), ABCB1 (forward: 5′-GTTCGCAACCCCAAGATCCT-3′ and reverse: 5′-CAATG GTGGTCCGACCTTTT-3′), and GAPDH endogenous control (forward: 5′-CTGAGA ACGGGAAGCTTGTCA-3′ and reverse: 5′-CC AGTGGACTCCACGACGTA-3′). The 20 μL reaction mixtures contained 10 μL of SYBR green PCR master mix (Applied Biosystems), 1 μL of cDNA template, 2 μL each of forward and reverse primers, and 5 μL of RNase-free water; cDNA derived from 5 ng of total RNA was used for each reaction. The reactions were done in triplicates using an ABI Prism 7000 Sequence Detection System (Applied Biosystems). The amplification of PXR, CYP3A4, ABCB1, and GAPDH was carried out as follows: 10-min initial denaturation at 95°C, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. The mRNA levels of PXR gene in the healthy liver tissues were expressed as the amount of target normalized to the endogenous control and relative to the PXR mRNA expression level in the liver tissue harboring the reference genotype at all loci.

Statistical analysis. The Fisher exact test was used to determine conformity with Hardy-Weinberg equilibrium and to assess genotype and allele distributions. χ² test was applied to compare interethnic variability among genotypic groups. The test of selective neutrality, based on either the infinite-allele model (Ewens-Watterson test) or the infinite-site model (Tajima's test), was used to test for sample homogeneity of each group in the Asian population. The nonparametric Mann-Whitney U test and Kruskal-Wallis test were used to assess differences in hepatic mRNA expression and association with haplotype structures. The effect of positive genotype-phenotype correlations was further explored by using nonparametric tests to examine the influence of haplotype structures on doxorubicin disposition in breast cancer patients. All statistical analysis and estimation of confidence intervals were carried out using STATA (STATA Statistical Software v7.0, Stata Corp.) and Prism v.4.00 (GraphPad Software, Inc.). The level of significance was set at P < 0.05 for all comparisons unless otherwise stated.

PXR pharmacogenetics in healthy Asian populations, liver tissues, and breast cancer patients.Table 1 summarizes the PXR genotype and allele frequencies in the healthy Chinese, Malay, and Indian populations; cancer patients; and liver tissues. Fisher's exact tests validated that all the PXR genotype frequencies conformed to Hardy-Weinberg equilibrium. The sample homogeneity was confirmed by the Ewens-Watterson test and Tajima's test (Chinese, Malays, and Indians; P > 0.05). A total of 16 single-nucleotide polymorphisms were identified from a complete screening of all nine exons and exon-intron boundaries, which spanned ∼40 kb of the PXR gene. There were three polymorphisms in the proximal 5′-region (−1570C>T, −566A>C, and −298G>A), four coding region polymorphisms (1437G>A, 1448G>A, 1792A>G, and 1944T>C), six intronic variants (IVS2+55A>G, IVS2+78A>G, IVS3+72T>G, IVS4+285G>A, IVS5−93A>G, and IVS6−17C>T), and three polymorphisms in the 3′-UTR region (2180A>G, 2617A>C, and 2654T>C). The allelic frequencies of several of the polymorphic variants in the PXR gene differed significantly among the Chinese, Malay, and Indian ethnic groups (Table 1). The coding region polymorphisms were confined to exon 9 in all three Asian ethnic groups. The genotype and allele frequencies of PXR polymorphisms among the healthy Chinese liver tissues and breast cancer patients were similar to the frequencies found among the healthy study subjects.

LD analysis. Pairwise LD analysis for the identified PXR polymorphisms in all three ethnic groups showed strong linkage between the following polymorphisms: −1570C>T with −566A>C and −298G>A (|D′| > 0.82); 1792A>G with 1944T>C (|D′| = 1); the intronic polymorphism IVS2+55A>G with IVS2+78A>G, IVS3+72T>G and IVS4+285G>A (|D′| > 0.82); and IVS6−17C>T with IVS5−93A>G (|D′| > 0.85). The 3′-UTR polymorphism 2617A>C was in complete linkage with 2654T>C polymorphism (|D′| = 1) across all three ethnic groups.

The pairwise LD matrix showed four LD blocks within the PXR gene in Chinese and Malays and three LD blocks in the Indian population (Fig. 1A-C). The LD blocks were highly similar in the Chinese and Malay populations. In both populations, block 1 was represented by two highly linked 5′-UTR polymorphisms, −566A>C and −298G>A (|D′| = 0.92), block 2 by the IVS2+55A>G and IVS2+78A>G intronic polymorphisms (|D′| = 1.00), and block 4 by the 3′-UTR 2617A>C and 2654T>C polymorphisms. Block 3 was also similar in the Chinese and Malay populations and differed only by the presence of the IVS5−93A>G polymorphism in the Chinese (Fig. 1A) and the 2180A>G polymorphism in the Malays (Fig. 1B). The LD between blocks 1, 2, and 3 in the Chinese were relatively weak (block 1 versus block 2, |D′| = 0.13; block 2 versus block 3, |D′| = 0.32), but blocks 3 and 4 were strongly linked (|D′| = 0.77). Similar LD patterns were observed in the Malay population (block 1 versus block 2, |D′| = 0.25; block 2 versus block 3, |D′| = 0.40). Strong LD between blocks 3 and 4 was also observed in the Malay population (|D′| = 0.74). Among Indians, block 1 was represented by the three intronic IVS2+55A>G, IVS2+78A>G, and IVS3+72T>G polymorphisms, whereas block 2 represented a combination of IVS5−93A>G, IVS6−17C>T, 1437G>A, 1448G>A, 1792A>G, 1944T>C, and 2180A>G polymorphisms. The genetic constitution of block 3 in Indians was identical to those observed in the Chinese and Malay populations and composed of the 3′-UTR 2617A>C and 2654T>C polymorphisms (Fig. 1C). The inter-block linkage was stronger in the Indian population (|D′| = 0.67 between blocks 1 and 2 and between blocks 2 and 3, respectively).

PXR haplotypes and network analysis. The inferred haplotypes across all ethnic groups are shown in Table 2. A total of 15 haplotypes were determined in the Chinese population, and four high frequency haplotypes (H1-C, H2-C, H3-C, and H4-C) accounted for a cumulative frequency of 62.2%. Among the 20 haplotypes inferred in Malay population, 7 were identified as high-frequency haplotypes (H1-M, H2-M, H3-M, H4-M, H5-M, H6-M, and H7-M) and had a cumulative frequency of 63.2%. There were 19 haplotypes identified in Indian population, and four high-frequency haplotypes (H1-I, H2-I, H3-I, and H4-I) constituted 52.7% of the total haplotype frequencies.

The relationships between haplotypes were further analyzed by calculating the haplotype network using the median-joining algorithm (17). The derived genealogic tree showing the relationship among the common haplotypes indicated the emergence of three distinct, highly divergent haplotype clusters in each Asian ethnic group, which were designated as PXR*1A, PXR*1B, and PXR*1C (Fig. 2). The PXR*1A, PXR*1B, and PXR*1C haplotype clusters were each distinguished by a similar set of two tagged polymorphisms in the three Asian ethnic groups (Fig. 2A-C). The 1792A>G and 1944T>C polymorphisms were representative of PXR*1A haplotype cluster, whereas PXR*1B was delineated by IVS6−17C>T and 2654T>C polymorphisms. The PXR*1C haplotype was tagged by the IVS2+55A>G and IVS2+78A>G polymorphisms. PXR*1A haplotype cluster were most common in the Indian population (52.4%) compared with the Chinese and Malay ethnic groups (37.2% and 29.4%, respectively). The proportion of PXR*1B haplotype cluster was relatively common in the Malay (35%) and Chinese (28.7%) populations compared with the Indian ethnic group (19.4%). The frequency of PXR*1C haplotype cluster was similar in all three Asian ethnic groups (Chinese, 26.6%; Malays, 20.3%; Indians, 22.5%). The individual haplotypes having frequency of <1% were grouped as “minor” and treated as unclassified in the network analysis.

PXR haplotype clusters and hepatic expression. The effects of PXR*1A, PXR*1B, and PXR*1C haplotype clusters on the hepatic mRNA expression of PXR were investigated in healthy liver tissues (n = 41) obtained from Chinese patients. The frequencies of PXR*1A, PXR*1B, and PXR*1C haplotype clusters in the liver tissues were 29%, 33%, and 38%, respectively. The PXR hepatic mRNA expression levels showed significant associations with PXR haplotype clusters (Fig. 3A). The mean mRNA expression level of PXR in liver tissues harboring the PXR*1B haplotypes was ∼4-fold lower compared with liver tissues carrying the *1A and *1C haplotypes [PXR*1A versus PXR*1B: RQ, 4.01 ± 1.11 versus 0.94 ± 0.24 (P = 0.015); PXR*1B versus PXR*1C: RQ, 0.94 ± 0.24 versus 3.5 ± 0.74 (P = 0.023)]. No significant difference was observed in hepatic PXR mRNA expression between liver tissues carrying the PXR*1A and PXR*1C haplotype clusters (P > 0.05).

The phenotypic effects of the different PXR haplotype clusters were investigated by examining their influence on the expression of their downstream target genes, CYP3A4 (Fig. 3B) and ABCB1 (Fig. 3C). Liver tissues carrying the PXR*1B haplotype cluster had significantly lower hepatic mRNA expression of CYP3A4 compared with liver tissues carrying the non-PXR*1B (*1A + *1C) group of haplotypes [PXR*1B versus non-PXR*1B: RQ, 1.73 ± 2.01 versus 4.15 ± 3.65 (P = 0.030)]. The hepatic mRNA expression of ABCB1 was similarly lower in liver tissues carrying the PXR*1B haplotype cluster compared with liver tissues carrying the non-PXR*1B haplotype cluster [PXR*1B versus non-PXR*1B: RQ, 0.81 ± 0.51 versus 1.25 ± 0.34 (P = 0.060)].

PXR haplotype clusters and doxorubicin clearance. The effects of PXR*1A, PXR*1B, and PXR*1C haplotype clusters on the clearance (CL) of doxorubicin were investigated in Asian breast cancer patients. The majority of the breast cancer patients belonged to the Chinese ethnic group (74%), followed by Malays (18%) and Indians (8%). The frequencies of the PXR*1A, PXR*1B, and PXR*1C haplotype clusters in the breast cancer patients were 40%, 27%, and 33%, respectively. Because the PXR haplotype clusters were associated with altered expression levels of its downstream target genes, their effect on the clearance of doxorubicin, which is an ABCB1 substrate, was investigated. The clearance of doxorubicin was significantly lower in patients harboring the PXR*1B haplotypes compared with patients carrying the non-PXR*1B (*1A and *1C) haplotypes [PXR*1B versus non-PXR*1B: CL/BSA (L h⁻¹ m⁻²), 20.84 (range, 8.68-29.24) versus 24.85 (range, 13.80-55.66), P = 0.022; Fig. 4].

PXR induces the expression of several hepatic and intestinal phase I enzymes (CYP2C9, CYP2B6, CYP3A7, CYP3A4, CYP24, CYP2C19, and CYP2C8), phase II conjugating enzymes (UGT1A1 and GSTs), and phase III drug transporters (efflux transporters such as ABCB1, ABCC2, and ABCC3 and uptake transporters such as SLC21A6; refs. 18–21). Characterizing PXR genetic variations and identifying functionally linked polymorphisms in different ethnic groups would provide valuable insights on interethnic and interindividual variations in disposition of several drugs that are known PXR ligands or substrates of its target genes.

In the present study, the PXR gene was highly polymorphic and showed significant interethnic variations among the Chinese, Malay, and Indian healthy populations. The allelic frequency of the 5′-UTR −1570C>T polymorphism was significantly higher among Indians (0.34) and was similar to reported frequencies in Dutch (0.36), Caucasian (0.39), and African American (0.32) populations (10, 22). The 5′-UTR −566A>C polymorphism, which was previously shown to be associated with significant up-regulation of ABCB1 and CYP3A4 mRNA levels in colon tumor specimens (11), was similar among the three Asian ethnic groups (Table 1) but lower than in other populations (0.40 in Caucasians, 0.70 in African Americans; refs. 10, 22). Taken together, these functional studies suggest that homozygosity for the variant allele at either loci may be associated with increased transcription of downstream PXR target genes.

More than 15 allelic variants in the DNA binding domain and ligand binding domain regions of the PXR gene have been known to have significant functional consequences, including aberrant DNA binding and alterations in transactivation and expression of downstream targets like ABCB1 and CYP3A4 (3, 11, 12). However, the allelic frequencies of these polymorphisms are low in different populations and were not detected in the present study. The exon 9 1437G>A polymorphism, which occurs at a high frequency among African Americans (0.36), was present at low frequency in all the three Asian ethnic groups (0.01), similar to the Japanese population (0.02; refs. 10, 11, 22). The 1792A>G and 1944T>C polymorphisms previously reported in the Dutch population (0.24; ref. 22) were present at a higher frequency among the Chinese (0.43), Malays (0.34), and Indians (0.59). Among the six intronic polymorphisms identified, the IVS4+285G>A variant was highly polymorphic in the Chinese (0.64), Malay (0.54), and Indian (0.60) populations and was similar to reports in the Dutch population (0.69; ref. 22). The allelic frequency of the 3′-UTR 2180A>G polymorphism displayed a 4-fold higher frequency in the Malay ethnic group (0.21) compared with Caucasians (0.05; ref. 10), whereas the completely linked 2617A>C and 2654T>C polymorphisms (|D′| = 1) were present at a 3-fold higher allele frequency among Chinese (0.46) compared with Caucasians (0.16; ref. 10).

The PXR gene is involved in the regulation of several downstream target genes involved in the disposition of several therapeutic agents. Previous studies in preclinical models of PXR-null mice and zebra fish had established the regulatory role of PXR on expression patterns of several downstream target genes involved in drug metabolism (CYP3A) and drug transport (ABCB1, ABCC2, ABCC3, and SLC21A1; refs. 23–25). In the present study, individual polymorphisms or haplotypes did not reveal any significant correlations with hepatic mRNA expression levels of PXR or its downstream target genes. To further clarify genotype-phenotype correlations, we sought to identify haplotype clusters that could influence PXR hepatic mRNA expression and, hence, its downstream target genes. Of the three PXR haplotype groups that were identified in the Asian population, the PXR*1A haplotype cluster was tagged by 1792A>G and 1944T>C polymorphisms, which were previously shown to be in strong linkage (22); however, there are no reports of functional influences of these polymorphisms. Homozygosity for the intronic IVS2+55A>G and IVS2+78A>G variants that tagged the PXR*1C was previously reported to be significantly associated with higher oral midazolam clearance in African Americans (13). However, this association was absent among the European Americans, suggesting that the observed association of the PXR*1C with increased PXR mRNA expression could be due to possible linkage of the IVS2+55A>G and IVS2+78A>G polymorphisms with other unidentified ethnically dependent functional polymorphisms. The linked IVS2+55A>G and IVS2+78A>G polymorphisms occur at a higher frequency among African Americans (0.89 and 0.65, respectively) and European Americans (0.64 and 0.28, respectively) when compared with the Chinese (0.41 and 0.45, respectively), Malay (0.49 and 0.46, respectively), and Indian (0.38 and 0.37, respectively) ethnic groups. The interethnic variations in frequencies of IVS2+55A>G and IVS2+78A>G polymorphisms and their varying association with drug clearance suggest the potentially important role of PXR*1C in influencing interethnic variations in drug disposition. PXR*1B was tagged by the 2654T>C and IVS6−17C>T polymorphisms, of which the 2654T>C polymorphism was previously shown to be associated with reduced intestinal ABCB1 expression whereas the IVS6−17C>T polymorphism was related to intestinal CYP3A inducibility (10).

The present study suggests that haplotype groups, rather than individual polymorphisms or haplotypes across the PXR gene, may be important in influencing the disposition of its ligands such as doxorubicin. Hepatic tissues harboring the non-PXR*1B (PXR*1A and PXR*1C) genetic constitution were associated with a significantly higher PXR mRNA expression levels as well as CYP3A4 and ABCB1 compared with PXR*1B hepatic tissues. The higher clearance of doxorubicin observed in the non-PXR*1B breast cancer patients can be hypothesized to be probably due to higher hepatic expression levels of ABCB1 in these patients that can potentially result in greater efflux. Because the allelic frequency PXR*1B haplotype was higher in the Chinese (28.7%) and Malay (35.0%) compared with the Indian (19.4%) ethnic groups, the phenotypic effect of PXR*1B on the disposition of its ligands that are substrates of CYP3A4 and ABCB1 may be different in these Asian ethnic groups. These findings also suggest that the therapeutic implications may be greater in the Chinese and Malay populations compared with the Indian ethnic groups. Further studies are required to validate these findings on the disposition of other PXR ligands in the Asian population as well as other ethnic groups.

Breast cancer patients are also routinely administered premedications such as dexamethasone, which is a known PXR inductive agent that can potentially lead to drug interactions, thereby altering the expression and activity of its downstream targets such as CYP3A4 and ABCB1 (26, 27), which may in turn affect the disposition characteristics of drugs that are substrates of CYP3A4 and ABCB1. In the present study, however, it is unlikely that differences in doxorubicin clearance observed between the different PXR haplotype cluster groups could have been due to differential inductive effect of dexamethasone, which was uniformly prescribed to all breast cancer patients.

The results of the present study differ from a recent study by Hor et al. (28) in which the investigators failed to show a relationship between PXR, CAR, HNF4α, or CYP3A5*3 polymorphisms and the pharmacokinetics of doxorubicin or docetaxel in Asian breast cancer patients. The reason for the discrepancy between their results and the present study is unclear. It is possible that the absence of genotypic-phenotypic associations in the study by Hor et al. (28) could have been due to lack of comprehensive analysis of the PXR gene polymorphisms. Furthermore, a large number of intronic polymorphisms were identified in the present study, which were mainly absent in their study. These intronic polymorphisms had also been reported previously in African American, Caucasian, Dutch, and Asian populations (11, 22).

PXR controls transcriptional events as a heterodimer with RXR by binding to distinct DNA response elements. Several other nuclear receptors that form heterodimer complexes with RXR also use the same binding elements, including vitamin D receptor (VDR; NR1I1) and the constitutive androstane receptor (CAR; NR1I3). It has been established that PXR overlaps functionally with CAR in terms of ligand binding and gene activation due to their structural homology (5). Recent studies indicate that the overlap in PXR and CAR target genes extends well beyond the members of the CYP3A, CYP2C, GSTs, UGTs, and the canalicular ABCC2 transporter (29). Functional polymorphisms have been identified for both PXR and CAR, but interethnic genetic variation in humans has not yet been well characterized (3, 30). Interethnic differences in PXR haplotype constitution as observed in the present report suggest that future studies should also take into consideration polymorphisms in other coregulators such as CAR while investigating for interethnic and interindividual variations in drug disposition related to PXR target genes.

In conclusion, significant differences were observed in the genotype and allele frequencies of PXR polymorphisms and their linkage patterns among the distinct Asian ethnic groups. The PXR*1B haplotypes, which occurred at a higher frequency in Chinese and Malays compared with Indians, were found to be associated with significantly decreased hepatic mRNA expression of PXR and also its downstream target genes, CYP3A4 and ABCB1. The PXR*1B haplotype was also found to be significantly associated with reduced clearance of doxorubicin, thus highlighting the important role of PXR pharmacogenetics in the disposition of putative drug substrates. Taken together, these findings suggest that genetic polymorphisms and specific haplotype clusters in the PXR nuclear receptor could have significant contributory roles in affecting interethnic variations in drug disposition in the Asian populations. Further studies in different population groups would be valuable to delineate the influence of PXR pharmacogenetics on the pharmacokinetics and pharmacodynamics of drugs that are substrates of PXR or its downstream target genes.

No potential conflicts of interest were disclosed.