Uridine 5′-diphosphoglucuronosyltransferase genetic polymorphisms and response to cancer chemotherapy

Abstract

Pharmacogenetics aims to elucidate how genetic variation affects the efficacy and side effects of drugs, with the ultimate goal of personalizing medicine. Clinical studies of the genetic variation in the uridine 5′-diphosphoglucuronosyltransferase gene have demonstrated how reduced-function allele variants can predict the risk of severe toxicity and help identify cancer patients who could benefit from reduced-dose schedules or alternative chemotherapy. Candidate polymorphisms have also been identified in vitro, although the functional consequences of these variants still need to be tested in the clinical setting. Future approaches in uridine 5′-diphosphoglucuronosyltransferase pharmacogenetics include genetic testing prior to drug treatment, genotype-directed dose-escalation studies, study of genetic variation at the haplotype level and genome-wide studies.

Cancer constitutes one of the main leading causes of death worldwide. One of the principal modalities of cancer treatment is chemotherapy. There is high interindividual variability in response to anticancer agents, some of which is caused by inherited variation in drug metabolizing enzymes. Pharmacogenetics investigates how genetic variation affects drug efficacy and side effects, with the goal of individualizing medical treatment to improve patient care.

Many anticancer agents are metabolized by glucuronidation, a metabolic reaction that increases drug solubility in water and facilitates their biliary or urinary excretion from the human body [1-3]. The glucuronidation reaction involves conjugation of uridine 5′-diphosphoglucuronic acid to aglycones that contain oxygen, carboxyl, nitrogen or sulfur functional groups via uridine 5′-diphosphoglucuronosyltransferase (UGT) enzymes. UGTs are differentially expressed throughout the human body. Most human UGTs are found in the liver (e.g., UGT1A1, UGT1A3, UGT1A4, UGT1A5, UGT1A6, UGT1A9, UGT2B4, UGT2B7, UGT2B10, UGT2B15 and UGT2B17), the main organ responsible for drug metabolism. However, other UGTs (e.g., UGT1A7, UGT1A8, UGT1A10, UGT2B11 and UGT2B28) are expressed exclusively in extrahepatic tissues [4-7]. The UGT genes are classified into families, UGT1 and UGT2, and, based on their sequence similarity, are further organized into subfamilies, UGT1A, UGT2A and UGT2B. Extensive genetic variation exists in the UGT1A and UGT2B genes [8,9]. Table 1 provides an overview of the many anticancer drugs that are glucuronidated in humans. Specific allele frequencies and a description of the in vitro effects of the UGT variants investigated in pharmacogenetic studies of anticancer agents are listed in Table 2.

Table 1. List of anticancer drugs metabolized by glucuronidation.

Drug	Class	UGT enzyme	Ref.
ABT-751	Antimitotic agent	UGT1A1, UGT1A4, UGT1A8 (major) and UGT2B7 [187]	[187-189]
Anastrozole	Aromatase inhibitor	UGT1A4 [190]	[190,191]
Dasatinib	Tyrosine kinase inhibitor	Unknown	[192]
Desmethyl arzoxifene and 4′F-desmethyl arzoxifene	Benzothiophene selective estrogen receptor modulators	Unknown	[193]
5,6-dimethylxanthenone-4- acetic acid	Vascular-disrupting agent	UGT1A9 and UGT2B7 (major) [194]	[194-197]
Doxorubicin	Anthracycline	Unknown	[198]
Epirubicin	Anthracycline	UGT2B7 [148]	[146-149,199-204]
Erlotinib	Tyrosine kinase inhibitor	Unknown	[205]
Etoposide	Topoisomerase II inhibitor	UGT1A1 [206,207]	[206,207]
Flavopiridol	Cyclin-dependent kinase inhibitor	UGT1A1, UGT1A4, UGT1A9 (major) and UGT1A10 [170,171]	[75,167,170,171]
Flutamide	Nonsteroidal anti-androgen drug	UGT1A1 and UGT1A6 (major) [208]	[208]
N-(2-hydroxyethyl)-3,5- dinitrobenzamide 2-mustard prodrugs: SN 27858 (PR-140A), SN 29546, SN 27686 and SN 29893	Dinitrobenzamide mustards	Unknown	[209]
Indisulam	Multitargeted cell-cycle inhibitor	Unknown	[210,211]
Lasofoxifene	Selective estrogen receptor modulator	UGT1A1, UGT1A3, UGT1A6, UGT1A8, UGT1A9 and UGT1A10 [212]	[212]
2-methoxyestradiol	Angiogenesis inhibitor	Unknown	[213]
NU/ICRF 505	Topoisomerase I inhibitor	UGT1A1, UGT1A8, UGT1A9 (main enzyme for tyrosine glucuronide formation), UGT1A10 and UGT2B7 (main enzyme for O-glucuronide formation) [214,215]	[214-216]
Raloxifene	Selective estrogen receptor modulator	UGT1A1 (main enzyme for 6-glucuronide formation), UGT1A3, UGT1A8, UGT1A9 (major enzyme for 4′-glucuronide formation), UGT1A10 and UGT2B7 [136-138]	[134, 136-139, 217,218]
SN-38 (active metabolite of irinotecan)	Topoisomerase I inhibitor	UGT1A1 (major), UGT1A3, UGT1A6, UGT1A7 and UGT1A9 [16-22,215]	[16-22,215]
Sorafenib	Tyrosine protein kinase inhibitor	UGT1A9 [304]	[304]
Tamoxifen	Anti-estrogen	UGT1A4 for N-glucuronidation of tamoxifen and 4-hydroxytamoxifen [112,114,115,119]; UGT1A1, UGT1A8 UGT1A9, UGT1A10, UGT2B7, UGT2B15 for O-glucuronidation of 4-hydroxytamoxifen and endoxifen [113,115,116]	[110-117,119,120]
TAS-103	Topoisomerase I and II inhibitor	UGT1A1 [174]	[174,175,219]
Tipifarnib	Farnesyltransferase inhibitor	Unknown	[220,221]
Topotecan	Topoisomerase I inhibitor	Unknown	[222]
Vorinostat	Histone deacetylase inhibitor	UGT1A1, UGT1A3, UGT1A7, UGT1A8, UGT1A9, UGT1A10, UGT2B7 and UGT2B17 (major) [164,165]	[164,165]

UGT: Uridine 5′-diphosphoglucuronosyltransferase.

Table 2. Allele frequencies and in vitro effects of UGT1A variants examined in pharmacogenetic studies of anticancer drugs.

UGT	Expression	UGT alleles	Nucleotide change (amino acid change)	In vitro effect	Frequency
UGT1A1	Hepatic	UGT1A1*1		Normal activity	61–72% in Caucasians, 45–52% in black populations, 90–98% in Japanese populations, 87–93% in Koreans and 84% in Chinese populations [9,56]
		UGT1A1*6	211G>A(G71R) [223;	Reduced activity [18,57]	0–1% in Caucasians, 0% in black populations, 13–18% in Japanese populations, 21–24% in Koreans and 23% in Chinese populations [9,56,67]
		UGT1A1*27	686C>A (P229Q) [223	Reduced activity [18]	0% in Caucasians and black populations and 1–4% in Asians [23,71]
		UGT1A1*28	A(TA)6TAA>A(TA)7TAA [28]	Reduced activity [16,28,224]	29–39% in Caucasians, 35–45% in black populations, 9–13% in Japanese populations, 7–13% in Koreans and 16% in Chinese populations [9,56]
		UGT1A1*60	−3279T>G [225]	Reduced activity [30,71]	44–55% in Caucasians, 85% in African–Americans, 17–26% in Japanese populations, 24–33% in Koreans and 30% in Chinese populations [30,56]
		UGT1A1*93	−3156G>A [225]	Reduced activity [30]	30–31% in Caucasians, 29% in African–Americans and 10% in Asians [30,68]
UGT1A4	Hepatic	UGT1A4*1a		Normal activity
		UGT1A4*2	70C>A (P24T) [186]	Substrate-dependent effect [186,226]	8% in Caucasians [186]
		UGT1A4*3b	142T>G (L48V) [186]	Decreased or no activity [186]	9% in Caucasians and 13% in Japanese populations [186,227]
UGT1A7	Extrahepatic	UGT1A7*1a		High activity	34–42% in Caucasians, 38% in African–Americans and 59% in Japanese populations [9]
		UGT1A7*2	387T>G/391C>A/392G>A (N129K/R131K) [23]	High activity [18,23]	24–34% in Caucasians, 39% in African–Americans, 15% in Japanese populations and 27% in Taiwanese Chinese populations [9,77]
		UGT1A7*3	387T>G/391C>A/392G>A/622T>C (N129K/R131K/W208R) [23]	Low activity [18,23]	23–36% in Caucasians, 23% in African–Americans, 26% in Japanese populations and 15% in Taiwanese Chinese populations [9,77]
		UGT1A7*4	622T>C (W208R) [23]	Low activity [18,23]	1–2% in Caucasians, 1% in black populations, 0% in Japanese populations and 0% in Taiwanese Chinese populations [9,77]
UGT1A8	Extrahepatic	UGT1A8*1a		Normal activity	55% in unknown population [228]
		UGT1A8*2	UGT1A8*2 (518G>C, A173G) [228]	Similar activity to UGT1A8*1a [228]	15% in unknown population [228]
		UGT1A8*3	UGT1A8*3 (830G>A, C277Y) [228,229]	Severely reduced activity [228,229]	2% in unknown population [228]
UGT1A9	Hepatic	UGT1A9*1a
		UGT1A9*1b	−118(dT)9>10 [72]	Luciferase activity increased in one study [72], but another study [74] did not find a significant change in luciferase activity or an association with hepatic UGT1A9 protein content	36–41% in Caucasians, 44% in black populations and 51–60% in Asians [68,72,75,230]
		I399C>T	I399C>T [73]	Increased protein expression and activity in one study [73]. No association with gene expression and activity in another study [75]	38–49% in Caucasians, 42% in Africans and 64% in Japanese populations [73,75,83]
UGT1A10	Extrahepatic	UGT1A10*1a		Normal activity
		UGT1A10*2a	415G>A (E139K) [231]		0% in Caucasians, 3–5% in black populations and 0% in Asians [231]
UGT2B7	Hepatic	UGT2B7*1a			46–51% in Caucasians and 73% in Asian–Americans and Japanese populations [9]
		UGT2B7*2a	802C>T (H268Y) [232]	No change in substrate activity for most substrates [148,232-234], increased activity for very few substrates [233]	49–54% in Caucasians and 27% in Asian–Americans and Japanese populations [9]
		−161T>C	−161T>C (relative to the ATG) [151]	Decreased activity [151]	49% in Caucasians and 69% in African–Americans [151]
UGT2B17	Hepatic	UGT2B17*1			68–69% in Caucasians and 79% in African–Americans [235,240]
		UGT2B17*2	Deletion of a 150-kbp region spanning the UGT2B17 gene [235,236]	Reduced activity in the homozygous state [237-241]	19–33% in Caucasians, 21% in African–Americans, 22% in Yorubans and 84% in Japanese populations and Chinese populations [235,239,240,242,243]

UGT: Uridine 5′-diphosphoglucuronosyltransferase.

This article will focus on how genetic variation in the UGT genes affects anticancer drug response, including both toxicity and efficacy. We report on anticancer drugs for which there is considerable glucuronidation and for which the UGTs responsible for the reaction have been identified. We will discuss pharmacogenetic studies of irinotecan, tamoxifen, raloxifene, epirubicin, vorinostat, flavopiridol and TAS-103 in detail. In this article, we use the nomenclature described in the UGT nomenclature webpage [301].

Irinotecan

The best example of how UGT genetic variation alters drug response is provided by irinotecan. This anticancer drug is used in many different schedules and disease settings, including colorectal cancer, small-cell lung cancer, breast cancer and gastric cancer [10-12]. It has also been approved by the US FDA for use in combination with 5-fluorouracil and leucovorin (folinic acid, fluorouracil and irinotecan [FOLFIRI]) for first-line treatment of patients with metastatic colorectal cancer and as second-line therapy for metastatic colorectal cancer refractory to 5-fluorouracil and leucovorin treatment [302].

Irinotecan is a prodrug of SN-38, a topoisomerase I inhibitor [13]. The cytotoxic activity of SN-38 is 100- to 1000-times greater than that of irinotecan [14]. SN-38 is further metabolized and deactivated in the liver to SN-38G [15], primarily by UGT1A1 [16-18] and to a lower extent also by UGT1A3, UGT1A6 and UGT1A9 [17-21]. UGT1A7 metabolizes SN-38 in vitro to a very high extent [18-20,22], but, since it is exclusively found in extrahepatic tissues [23] (reviewed in [4,24,25]), it does not contribute to SN-38 glucuronidation in the liver, where most SN-38G is formed.

Patients receiving irinotecan commonly experience severe myelosuppression (resulting in neutropenia) and delayed-onset diarrhea. Severe neutropenia and diarrhea seem to be caused by SN-38-induced damage to the bone marrow and gastrointestinal mucosa, respectively. Both myelosuppression and diarrhea need to be closely monitored. When experienced as grade 3 or 4 toxicities, they require treatment delay to allow for recovery and a decrease in dosing before the next irinotecan treatment.

More than 35 pharmacogenetic studies have evaluated whether there is a genetic predisposition to the risk of severe toxicity from irinotecan. Most studies have focused on investigating how genetic variation in the UGT1A1 gene affects the drug pharmacodynamics owing to its primary role in the elimination of SN-38. An association between irinotecan toxicity and impaired UGT1A1 activity was initially observed in cancer patients suffering from Gilbert’s syndrome [26,27], a common and mild inherited liver disorder that causes mild hyperbilirubinemia owing to a deficiency in UGT1A1 activity [28]. Case reports indicated that these patients developed severe diarrhea, neutropenia and transient increases in unconjugated bilirubin following irinotecan treatment [26,27]. The patients had high biliary SN-38 concentrations, suggesting reduced glucuronidation. Around the same time, in vitro studies showed that SN-38 is metabolized by UGT1A1, and there is an inverse correlation between the number of thymineadenine (TA) repeats in the promoter region of UGT1A1 and glucuronidation of SN-38 and bilirubin [16,29]. Since then, many clinical studies have investigated whether genetic polymorphisms in the UGT1A genes involved in SN-38 disposition, particularly UGT1A1, predispose patients to severe irinotecan toxicity. These studies have been summarized in Table 3.

Table 3. Polymorphic variation in UGT associated with irinotecan toxicity (p < 0.05).

UGT allele	Clinical effect
UGT1A1*28	Increased severe neutropenia in Caucasians [32,34-38,46,88] and Japanese populations [41,42] Increased leukopenia in Japanese populations [39-41] Decreased ANC nadir in Caucasians [31,32] Increased severe diarrhea in Caucasians [38,43-46] and Japanese populations [39,40]
UGT1A1*60	G/G genotype was associated with severe hematologic toxicity when compared with T/T at frst cycle in Caucasians [70] but was not confirmed in multivariate analysis
UGT1A1*93	Increased severe neutropenia in Caucasians [32,37,69] Decreased ANC nadir in Caucasians [32]
UGT1A1*6	Increased severe neutropenia in Japanese populations [41,42,48,49,60,64], Koreans [48,49] and Chinese populations [59] Increased leukopenia in Japanese populations [41]
UGT1A7*2	Lack of severe neutropenia or diarrhea in Caucasians [47]
UGT1A7*3	Lack of severe neutropenia or diarrhea in Caucasians [47] Increased severe hematologic toxicity in Caucasians [70] Increased severe diarrhea in Koreans [48,49]
UGT1A7*4	Increased severe neutropenia and/or diarrhea in Caucasians [46]
UGT1A9*1b	Increased severe neutropenia or diarrhea in Caucasians [47] Increased severe neutropenia in Japanese populations [60] Severe hematologic toxicity in Caucasians [70] Decreased severe diarrhea in Koreans [48,49]

ANC: Absolute neutrophil count; UGT: Uridine 5′-diphosphoglucuronosyltransferase.

UGT1A variation & irinotecan toxicity

Many irinotecan pharmacogenetic studies have investigated whether the UGT1A1*28 allele is associated with the risk of developing severe drug-related toxicity. UGT1A1*28 has an additional TA repeat [A(TA)₇TAA] in the TATA box of the promoter region, which in most individuals contains the A(TA)₆TAA allele known as UGT1A1*1. Transcriptional activity is inversely related to the number of TA repeats, and as the number of TA repeat increases, UGT1A1 glucuronidation activity is significantly reduced [28-32]. UGT1A1*28 is also associated with reduced UGT1A1 mRNA expression [33]. The first prospective irinotecan pharmacogenetic study investigated whether the incidence of toxicity correlated with variability in the UGT1A1 promoter. The trial enrolled predominantly Caucasian patients with refractory disease to be treated with 300–350 mg/m² irinotecan once every 3 weeks. At 300 mg/m², patients with one or two UGT1A1*28 alleles had a trend toward lower absolute neutrophil counts (n = 20) [30]. In 65 patients enrolled in the same trial and treated with 350 mg/m² irinotecan (the current dosage of irinotecan, after a change in FDA recommendations), 50% of UGT1A1*28 homozygotes experienced grade 4 neutropenia as the most common toxicity compared with 13% of heterozygotes and 0% of the UGT1A1*1/*1 patients [32]. Having a UGT1A1*28/*28 genotype conferred a 9.3-fold risk (95% CI: 2.4–36.4) of suffering neutropenia when compared with UGT1A1*28/*1 and UGT1A1*1/*1. Multiple subsequent irinotecan studies have also reported increased risk in carriers of UGT1A1*28/*28 for developing neutropenia [34-38]. The largest study group was trial N9741, comprised of 520 advanced colorectal cancer patients treated with combination therapy (irinotecan/5-fluorouracil/leucovorin: n = 114; oxaliplatin/5-fluorouracil/leucovorin: n = 299; and irinotecan/oxaliplatin: n = 107) [35]. Preliminary reports found a statistically significant association between UGT1A1*28/*28 and higher incidence of severe neutropenia in all patients combined and in patients on the irinotecan/oxaliplatin arm. Irinotecan/oxaliplatin patients homozygous for UGT1A1*28 had a 15.3-fold (95% CI: 3.0–77.9) and 35.0-fold (95% CI: 3.6–40.9) increased risk of developing grade 4 neutropenia and grade 3 febrile neutropenia, respectively. Japanese carriers of UGT1A1*28 are also at higher risk for developing severe leukopenia and neutropenia [39-42]. For example, in a Japanese cohort of 118 patients in different regimens containing irinotecan, the UGT1A1*28 allele distribution in patients who experienced grade 4 leukopenia was significantly different from that of patients without toxicity [39]. Multivariate analysis also showed that patients with one or two UGT1A1*28 alleles had a significant risk for severe toxicity.

An association between the incidence of diarrhea and UGT1A1*28 has also been observed by some investigators. In a study of Caucasians receiving irinotecan-containing regimes, 70% of the carriers of UGT1A1*28/*28 developed diarrhea versus 33% of UGT1A1*28/*1 versus 17% of UGT1A1*1/*1 [43]. Metastatic colorectal patients on FOLFIRI who were homozygous for UGT1A1*28 were also at higher risk of severe diarrhea [38]. The presence of the UGT1A1*28 allele also correlated with severe diarrhea and was a significant risk factor for toxicity in additional studies in Caucasian and Japanese populations [39-40,44-46]. However, other studies have not provided evidence of an association between severe irinotecan-induced diarrhea and UGT1A1*28, possibly owing to insufficient sample size, low allele frequency in Asian studies, lower irinotecan doses and heterogeneity in treatment schedules [42,47-54].

To investigate the reason for the inconsistencies observed in the relationship between UGT1A1*28 and severe irinotecan-induced toxicity, a meta-analysis was conducted integrating the results of nine clinical trials including 821 patients in total and encompassing different dose schedules (weekly, biweekly and every 3 weeks) [55]. The study found that the risk for severe hematologic toxicity in patients with UGT1A1*28/*28 was dose dependent. The risk of developing grade 3–4 neutropenia in patients with UGT1A1*28/*28 genotype is significantly higher than for carriers of UGT1A1*1/*28 or UGT1A1*1/*1 at high (dose: >250 mg/m²; odds ratio [OR]: 3.22; 95% CI: 1.52–6.81; p = 0.008) and medium doses (dose: 150–250 mg/m²; OR: 27.8; 95% CI: 4.00–195; p = 0.005) but not at lower doses (dose: <150 mg/m²; OR: 1.80; 95% CI: 0.37–8.84; p = 0.41) [55]. The authors concluded that the UGT1A1*28/*28 genotype may be useful as a predictive marker of toxicity only at intermediate or high irinotecan doses (150–250 mg/m²) administered every 2 or 3 weeks, but treatment decisions at lower doses do not need to be made based on genotype. No associations were found between UGT1A1*28 and diarrhea, and for incidence of diarrhea in UGT1A1*28/*28 patients and dose (p = 0.8).

UGT1A1*6 (211G>A, G71R), a missense mutation found in Asians [9,56], is associated with reduced enzyme activity [18,57,58]. It has been shown to predict for toxicity in Asian patients receiving irinotecan for the treatment of non-small-cell lung cancer. A clinical trial of 81 Koreans given irinotecan and cisplatin found an association between homozygosity for UGT1A1*6 and the incidence of grade 4 neutropenia (OR: 6.11) [48,49]. Additional studies conducted in Japanese and Chinese have confirmed the association between UGT1A1*6 and incidence of severe leukopenia and neutropenia [41,42,59,60]. Coexistence of UGT1A1*28 and UGT1A1*6 has been shown to decrease glucuronidation activity in an additive manner [39,41,61-64] and genotyping both alleles would be important to predict toxicity in Asians (reviewed in [65,66]) Genotyping UGT1A1*6 in Caucasians, however, has no practical value due to its rarity [32,34].

Irinotecan pharmacogenetic studies have also investigated the impact of UGT1A1*27, UGT1A1*93 and UGT1A1*60 on toxicity. The effect of the UGT1A1*27 (686C>A, P229Q) allele, which reduces glucuronidation activity, has not been evaluated in Caucasians due to its rarity [32,34]. It has been found to be associated with severe neutropenia and diarrhea in Japanese but as it co-occurs with UGT1A1*28, it is hard to assess its effect [39,62,67]. UGT1A1*93 (−3156G>A) and UGT1A1*60 (−3279T>G) occur in Caucasians, Asians and Africans [30,56,68]. UGT1A1*93 was strongly correlated with severe neutropenia in irinotecan patients with advanced disease refractory to other agents and seemed to distinguish between different phenotypes of total bilirubin (an established marker of UGT1A1 status) better than UGT1A1*28 [32]. Consistent with this hypothesis, a study of colorectal cancer patients treated with FOLFIRI demonstrated a significant trend of increased incidence of severe neutropenia among −3156 genotypes [37]. Severe hematologic toxicity was more frequent in individuals with −3156A regardless of whether they carried the UGT1A1*1 or UGT1A1*28 alleles. A hazard ratio for severe hematological toxicity of 8.4 was found for patients with −3156A/A genotype compared with −3156G/G. Homozygosity for −3156G>A was also related to the risk of neutropenia in irinotecan patients with extensive-stage small-cell lung cancer [69]. However, as no functional studies have been performed for UGT1A1*93, and since this allele is highly linked to UGT1A1*28 and UGT1A1*60 in Caucasians and Asians [32,62,67,70], further studies are needed. UGT1A1*60 has decreased transcriptional activity [71]. In Caucasian metastatic patients treated with FOLFIRI, UGT1A1*60/*60 was associated with severe hematologic toxicity in the first treatment cycle when compared with −3279T/T carriers, although it was not a significant predictor after multivariate analysis [70]. In Japanese patients, UGT1A1*60 has been associated with severe leukopenia and/or diarrhea [40]. Homozygosity for UGT1A1*60 also increased the risk of severe toxicity. However, in multivariate analyses including UGT1A1*28, this association did not hold, although the association with UGT1A1*28 did, suggesting that UGT1A1*28 is a stronger predictive factor for toxicity than UGT1A1*60. Another study investigating the role of UGT1A1*60 in the severe side effects experienced by Korean patients could not demonstrate an association with toxicity [48]. Owing to the high linkage between UGT1A1*28, UGT1A1*60 and UGT1A1*93, more evidence is needed to assess whether genotyping UGT1A1*60 and UGT1A1*93 would add predictive power to that provided by UGT1A1*28, especially in Caucasian populations where UGT1A1*28 is common.

UGT1A9 and UGT1A7 also metabolize SN-38. UGT1A9 is expressed in the liver but UGT1A7 is only expressed in extrahepatic sites. Common polymorphisms in UGT1A9 include UGT1A9*1b (−118(dT)_9>10), generated by insertion of an extra thymidine in the promoter region [72], and I399C>T [73]. The functional significance of both polymorphisms is unclear. Although UGT1A9*1b has been associated with increased luciferase activity [72], further studies have not shown a significant increase in reporter gene expression or alterations in hepatic protein expression or activity when compared with the reference allele UGT1A9*1a [73-75]. I399C>T was originally correlated with increased UGT1A9 protein levels and activities and in low linkage disequilibrium with UGT1A9*1b in Caucasian livers [73]. However, a second study did not find an association between I399C>T and UGT1A9 mRNA expression and activity, and found I399C>T and UGT1A9*1b to be in complete linkage disequilibrium [75]. An in vivo study also could not find an association between I399C>T and glucuronidation of mycophenolic acid, a probe UGT1A9 substrate [76]. UGT1A7 is also polymorphic; UGT1A7*2 (387T>G/391C>A/392G>A, N129K/R131K) and UGT1A7*3 (387T>G/391C>A/392G>A/622T>C, N129K/R131K/W208R) are common while UGT1A7*4 (622T>C, W208R) is very rare (reviewed in [9,77]). UGT1A7*2 has similar glucuronidation activity to UGT1A7*1. Both UGT1A7*3 and UGT1A7*4 have reduced activity [18,23].

Most studies of the effect of genetic variation in UGT1A9 and UGT1A7 on irinotecan toxicity have shown an association between low-activity alleles and increased toxicity. In Koreans, increased incidence of grade 3 diarrhea has correlated with both UGT1A9*1a/*1a and UGT1A7*3/*3 [48]. In this study, increased incidence of severe toxicity was also found in the presence of combined mutations with low activity (UGT1A1*28 and UGT1A7*4). However, studies in Japanese and Caucasians found that severe hematologic toxicity was more frequent in individuals with UGT1A9*1b [60,70], although the results did not hold in multivariate analyses [70]. Two Caucasian clinical trials showed that carriers of the low-activity allele UGT1A7*3 had increased severe hematologic toxicity and/or diarrhea [46,70]. UGT1A7 −57T>G, in linkage disequilibrium with UGT1A7*4, was associated with anemia, and both leukopenia and thrombocytopenia were more prevalent in patients with the high-risk alleles UGT1A1*28, UGT1A7*2 and UGT1A7 −57T>G [78]. Overall, these results do not allow any firm conclusion to be drawn owing to the high linkage existing between UGT1A7*3 and UGT1A9*1a with both UGT1A1*28 and UGT1A1*6 [47,48,70,79-82]. Regarding UGT1A9 I399C>T, no association was found between this allele and severe neutropenia in Japanese patients [83]. In this same study, this variant was in linkage with UGT1A9*1b, UGT1A7*2 and UGT1A7*3, and 85% of the T alleles were linked with the UGT1A1 wild-type haplotype.

The most common UGT1A1 haplotype in Caucasians (frequency of 34.2%) harbors all reference sequence alleles except for UGT1A9*1a: UGT1A1 −3279T, UGT1A1 −3156G, UGT1A1 −53(TA)₆, UGT1A7 +387T, UGT1A7 +622T and UGT1A9 −118(T)₁₀. In a clinical study, this haplotype together with sex had a protective effect on irinotecan-induced severe hematologic toxicity during the entire course of therapy [70]. In Japanese patients, haplotypes containing UGT1A1*28, UGT1A7*1 and UGT1A9*1b or UGT1A1*6, UGT1A7*3 and UGT1A9*1a had a greater incidence of severe neutropenia when compared with the reference sequence haplotype containing UGT1A1*1, UGT1A7*1 and UGT1A9*1a, demonstrating the clinical impact of UGT1A1*28 and UGT1A1*6 [42].

UGT1A1 variation & irinotecan pharmacokinetics

Many studies have used the relative extent of SN-38 glucuronidation (area under curve [AUC]_SN-38G/AUC_SN-38) as a marker for UGT1A1 activity. The majority of the studies investigating the role of UGT1A1*28 on irinotecan pharmacokinetics have found a significant reduction in the relative extent of glucuronidation in Caucasian [31,32,36,52,54,62,84,85] and Japanese subjects [40-42] carrying this allele. A similar relationship has been found between UGT1A1*6 and the extent of glucuronidation in Japanese patients [41-42,62,63], Koreans [48,49], Chinese patients, Malays and Indians [59]. An additive effect of UGT1A1*28 and UGT1A1*6 on the relative extent of SN-38 glucuronidation has been reported in Japanese [42]. Caucasian and Japanese carriers of UGT1A1*60 have also shown decreased AUC_SN-38G:AUC_SN-38 ratios [40,42,70].

UGT1A7*3 and homozygosity for UGT1A9*1a/*1a have been reported to be associated with reduced glucuronidation ratios in Koreans [48,49]. The same association with UGT1A7*3 was observed in Caucasians [70]. As both UGT1A7*3 and UGT1A9*1a are linked with UGT1A1*28 and UGT1A1*6 [42,47,81,82,86], and UGT1A7 is not expressed in the liver, the effect of these variants on SN-38 glucuronidation requires further investigation. UGT1A9 I399T carriers have shown increased glucuronidation ratios in Japanese populations, but this is very likely to be due to the close association of I399C with UGT1A1*28, UGT1A1*6 or UGT1A1*60, as after stratifying patients by UGT1A1*28, UGT1A1*6 or UGT1A1*60 haplotypes associated with reduced glucuronidation activity, the significant effect was no longer observed [83]. In another study, I399T alleles were associated with higher glucuronidation activity and lower systemic exposure to SN-38 when analysis was restricted to patients with UGT1A1*1/*1 genotype [87]. However, when analyzing only UGT1A1*6/*6 patients, the effect was not observed. In this population, there was very low degree of linkage between I399C>T and the rest of the UGT1A1 functional variants (i.e., UGT1A1*28, UGT1A1*6, UGT1A1*93 and UGT1A1*60) and only a weak linkage with UGT1A9*1b.

UGT1A1*28 has also been associated with increased SN-38 AUC in some studies [31,32], but it does not have an effect on irinotecan AUC, as expected. In a trial of Asian cancer patients (Chinese, Malays and Indians), UGT1A1*6 was associated with higher SN-38 exposure, but the same effect was not observed in UGT1A1*28 carriers [59]. In Koreans, patients with UGT1A1*6/*6, UGT1A7*3/*3 and UGT1A9*1a/*1a genotypes had increased SN-38 AUC [48]. Overall, these results suggest that the relative extent of glucuronidation of SN-38 is a good marker of UGT1A1 status.

A few studies in Japanese populations have analyzed the correlation between haplotypes and glucuronidation activity. Japanese patients bearing haplotypes harboring either UGT1A1*6 or UGT1A1*28 exhibited lower glucuronidation ratios than those without these alleles [42,62]. An additive effect of haplotypes containing UGT1A1*28 and UGT1A1*6 on reduced AUC ratio was also observed [42]. Haplotypes with UGT1A1*60 had reduced glucuronidation although the trend did not reach statistical significance. In another trial, reduced glucuronidation was observed in patients homozygous for the haplotype UGT1A1*6 UGT1A7 −57G UGT1A7*3 UGT1A9*1a (present in 15% of the patients) due mainly to the presence of UGT1A1*6 [82].

UGT1A variation & antitumor efficacy

Studies evaluating the role of UGT1A1*28 in antitumor response of Caucasian patients treated with irinotecan have given contradictory results. UGT1A1*28/*28 was associated with complete and partial response [70] and higher response rate [36] in FOLFIRI trials. However, other studies have found homozygosity for UGT1A1*28 to be associated instead with lower response [38] or to have no relationship with it [35,47,88]. Studies have reported an association between homozygosity for UGT1A1*28 and both stable disease [70] and decreased risk of tumor progression [36]. By contrast, another study found no significant improvement in time to progression in carriers of UGT1A1*28 [35,51].

Most studies have found no correlation between UGT1A1*28 and median survival [51], overall survival [35,70] or disease-free survival [88]. A marginally significant association has been found between poorer overall survival and UGT1A1*28 [43], and between lower median survival and UGT1A1*28/*28 [51]. However, another study observed a marginally significant tendency for better disease-free survival for UGT1A1*28/*28 carriers [37].

UGT1A1*6/*6 has been correlated with lower tumor response, progression-free survival and overall survival in non-small-cell lung cancer Asian patients [48,49] UGT1A1*93/*93, which also reduces glucuronidating activity, has been associated with partial and complete response in Caucasians, in agreement with the observations previously described with UGT1A1*28/*28 carriers [70]. However, a similar association was not observed in UGT1A1*60/*60 patients [70].

In metastatic Caucasian cancer patients, patients with UGT1A1*93/*93 genotype showed significantly better responses than those with UGT1A1*1/*1 [70]. In the same study, no associations with tumor response were observed with UGT1A1*60, UGT1A7*2, UGT1A7*3, UGT1A7*4 and UGT1A9*1b variants. In other studies, UGT1A9*1b/*1b was predictive of worse progression-free survival in Koreans [49], and UGT1A9*1a/*1a was associated with increased response in Caucasians [47].

Given the contradictory nature of some of these results, prospective and randomized studies are needed to evaluate whether UGT1A variation can predict antitumor efficacy, also considering additional factors, such as patient characteristics, the biological state of the tumors and environmental factors.

Genetic testing for irinotecan treatment

The irinotecan label was amended in 2005, prompted by findings from clinical trials linking UGT1A1*28 homozygosity to increased risk for neutropenia. The revised package insert recommends that patients with UGT1A1*28/*28 genotype should receive a lower irinotecan dose [302]. To facilitate the clinical integration of the genetic test, the FDA approved the Invader® UGT1A1 Molecular Assay from Third Wave Technologies, Inc. (WI, USA) [303]. This genetic test helps identify cancer patients who might be at increased risk of severe toxicity and could benefit from either an irinotecan dose reduction or alternative chemotherapy regimens [65].

Further evaluation of the relationship between genotype and severe toxicity has shown that UGT1A1*28/*28 may be useful as a predictive marker of toxicity at only intermediate or high irinotecan doses (150–250 mg/m²) administered every 2 or 3 weeks [55]. Patients with UGT1A*28/*28 genotype do not tolerate standard doses of irinotecan owing to excessive toxicity and should undergo dose reductions [32,34,43,89], the extent of which is still not established. Patients of Asian origin would benefit from genetic testing for UGT1A1*6 because of its predictive value for toxicity [90,91].

Genotype-directed dose-escalation studies

It can be postulated that the dose of irinotecan is suboptimal in patients with both UGT1A1*1/*28 and UGT1A1*1/*1 genotypes. These individuals may benefit from higher doses.

A recent study performed in FOLFIRI patients of Caucasian origin demonstrated the feasibility of optimization of irinotecan dosing according to UGT1A1*28 genotype [92]. By performing dose-escalation studies in patients with UGT1A1*1/*1 and UGT1A1*1/*28 genotypes, the clinical trial demonstrated that UGT1A1*1/*1 and UGT1A1*1/*28 patients could be safely treated every 2 weeks with irinotecan doses of 370 and 310 mg/m², respectively. These doses are considerably higher that the recommended irinotecan dose of 180 mg/m² in FOLFIRI, and demonstrates that patients who are not homozygous for UGT1A1*28 can tolerate higher doses of irinotecan. Additionally, a genotype-directed dose-escalation study of a population consisting mainly of Caucasian patients is ongoing and aims to determine the optimal doses of irinotecan administered as single agent and stratified by UGT1A1 genotype [93]. So far, the study shows that UGT1A1*1/*1 patients can be treated safely at higher doses and that dose escalation beyond the standard irinotecan dose is not tolerated by UGT1A1*1/*28 patients. Genotype-directed dosing, including UGT1A1*6 in addition to UGT1A*28, has also been studied. Korean patients treated with irinotecan and a fixed dose of capecitabine every 3 weeks can tolerate 350 mg/m² irinotecan if they have UGT1A1*1/*1, UGT1A1*28/*1 or UGT1A1*1/*6 genotypes, and 200 mg/m² if their genotypes are UGT1A1*28/*28, UGT1A1*28/*6 or UGT1A1*6/*6 [94]. Japanese patients harboring UGT1A1*28/*28, UGT1A1*28/*6 or UGT1A1*6/*6 genotypes tolerate lower irinotecan doses administered every 2 weeks (maximum tolerated dose: 150 mg/m²) than patients with one or two copies of the UGT1A1*1 allele [95].

Tamoxifen

Tamoxifen is an oral, nonsteroidal anti-estrogen for the prevention and treatment of steroid hormone receptor-positive breast cancer and for lowering breast cancer incidence in high-risk women [96-98]. The drug binds to the estrogen receptor and competitively inhibits the binding of estrogen in breast tissue. Tamoxifen undergoes extensive hepatic metabolism by cytochrome P (CYP) 450 [99-109], UGT [110-120] and sulfotransferase enzymes [116,121].

The most abundant tamoxifen metabolite is N-desmethyltamoxifen, produced by CYP3A4 and CYP3A5 [101-103,108,122]. Tamoxifen has two clinically active metabolites: trans-4-hydroxytamoxifen [123,124] and trans-4-hydroxy-N-desmethyl-tamoxifen [96,109,110,123-126], also known as endoxifen [127]. Trans-4-hydroxytamoxifen is a primary metabolite formed mainly by CYP2D6 [99,101,103-107,109], while endoxifen is formed by 4-hydroxylation of N-desmethyltamoxifen by CYP2D6 and by demethylation of trans-4-hydroxytamoxifen by CYP3A4 and CYP3A5 [101,102]. Endoxifen has equivalent potency to 4-hydroxytamoxifen [109,126], but as the patient plasma levels of endoxifen are higher than those of 4-hydroxytamoxifen [109,127], endoxifen may contribute more significantly to the anti-estrogenic action of tamoxifen, especially in CYP2D6-extensive metabolizers [109,123,126].

Tamoxifen and its two clinically active metabolites, trans-4-hydroxytamoxifen and trans-endoxifen, undergo extensive glucuronidation in humans [110-117,119,120]. Both tamoxifen and trans-4-hydroxytamoxifen undergo N-glucuronidation by UGT1A4 to form trans-tamoxifen-N⁺-glucuronide, trans-4-hydroxytamoxifen-N⁺-glucuronide and the geometrical isomer cis-4-hydroxytamoxifen-N⁺-glucuronide [113,115,116,120]. The isomerization reaction favors trans-4-hydroxytamoxifen over the cis-isomer [117,128]. Trans-4-hydroxytamoxifen and trans-endoxifen are conjugated by O-glucuronidation to form trans-4-hydroxytamoxifen-O-glucuronide by UGT2B7 [113,115], cis-4-hydroxytamoxifen-O-glucuronide by UGT1A10, UGT1A1, UGT2B7 and UGT2B15 [113,115,116], trans-endoxifen-O-glucuronide by UGT1A10, UGT1A8 and UGT2B7 [113], and cis-endoxifen-O-glucuronide by UGT1A10, UGT1A9 and UGT1A1 [113].

In vitro studies have investigated the effect of genetic variants in UGT1A1 (UGT1A1*28), UGT1A4 (UGT1A4*2 and UGT1A4*3b), UGT1A8 (UGT1A8*2 and UGT1A8*3), UGT1A10 (UGT1A10*2a) and UGT2B7 (UGT2B7*2a) on the glucuronidation of tamoxifen and its active metabolites (Table 4). UGT2B7*2a exhibited a decrease in glucuronidation activity against trans-4-hydroxytamoxifen and trans-endoxifen in human liver microsomes and cell homogenates [119]. Similar results were obtained for homozygotes compared with wild-type in human liver microsomes. UGT1A8*3 completely abolished the glucuronidation activity of both compounds, while UGT1A8*2 had a very minor effect (reduction) on endoxifen glucuronidation. UGT1A4*3b, a low-activity allele, showed increased activity against N-glucuronidation of tamoxifen and 4-hydroxytamoxifen in one study [129], and no effect in another [119]. The other variants had no effect compared with the reference alleles. Future studies should explore whether these polymorphisms can predict clinical response to treatment with tamoxifen.

Table 4. Pharmacogenetic studies of anticancer drugs that undergo glucuronidation.

Anticancer agent	Type of study	UGT allele or genotype	Effect
Tamoxifen	In vitro studies using UGT1A4-, UGT1A8- , UGT1A10- and UGT2B7-overexpressing HK293 cells [118,129] and human liver microsomes [118]	UGT1A1*28/*28	Nonsignificant decreases in O-glucuronidation activity against the trans isomers of 4-hydroxytamoxifen and endoxifen in human liver microsomes with genotypes UGT1A1*28/*28 compared with UGT1A1*1/*1, or UGT1A1*28/*28 and UGT1A1*28/*1 compared with UGT1A1*1/*1 [118]
		UGT1A4*2	No differences in N-glucuronidation activity against tamoxifen, trans-4-hydroxytamoxifen and cis-4-hydroxytamoxifen compared with UGT1A4*1a [129]. No significant association with N-glucuronidation activity of trans-4-hydroxytamoxifen in human liver microsomes [118]
		UGT1A4*3b	Increased activity against N-glucuronidation activity of tamoxifen, trans-4-hydroxytamoxifen and cis-4-hydroxytamoxifen [129]. No significant association with N-glucuronidation activity of trans-4-hydroxytamoxifen in human liver microsomes [118]
		UGT1A8*2	No difference in O-glucuronidation activity against trans-4-hydroxytamoxifen. Small but significant decrease against trans-endoxifen compared with wild-type UGT1A8*1a using cell homogenates [118]
		UGT1A8*3	No O-glucuronidation activity activity against trans-4-hydroxytamoxifen and trans-endoxifen using cell homogenates [118]
		UGT1A10*2a	No difference in O-glucuronidation activity activity against trans-4-hydroxytamoxifen and trans-endoxifen nusing cell homogenates [118]
		UGT2B7*2a	Decreased O-glucuronidation activity against trans-4-hydroxytamoxifen and trans-endoxifen in human liver microsomes and cell homogenates [118]
		UGT2B7*1a/*2a	Nonsignificant decrease of O-glucuronidation against trans-4-hydroxytamoxifen and trans-endoxifen compared with UGT2B7*1a/*1a in human liver microsomes [118]
		UGT2B7*2a/*2a	Significant decrease in of O-glucuronidation against trans-4-hydroxytamoxifen and trans-endoxifen glucuronidation activity compared with UGT2B7*1a/*1a in human liver microsomes [118]
Raloxifene	Study of postmenopausal females treated for osteoporosis [139]	UGT1A1*28	Increased glucuronide levels in UGT1A1*28/*28 carriers compared with UGT1A1*28/*1 or UGT1A1*1/*1. The parent drug concentrations were also increased in individuals with UGT1A1*28/*28 but not significantly [139]
Epirubicin	In vitro study in HEK-293 cell membranes expressing UGT2B7 encoded by UGT2B7*1a and UGT2B7*2a alleles [148]	UGT2B7*2a	No detectable differences in activity compared with UGT2B7*1a [148]
	In vitro study with human livers [149]	UGT2B7 haplotype 4†	Increased enzyme activity and gene expression. Diplotypes containing haplotype 4 had a significant 27% average increase in glucuronidation and more than a fivefold increase in mRNA expression compared with diplotypes without haplotype 4 [149]
	Clinical trial in breast cancer patients receiving adjuvant or neoadjuvant FEC100 every 3 weeks [155]	−161T>C	Reduced clearance and increased severe leukopenia [155]
Flavopiridol	Phase 1 clinical trial of patients with refractory neoplasms taking flavopiridol 1-h intravenous infusion daily [172]	UGT1A1*28	No association with pharmacokinetics or the occurrence and severity of diarrhea and neutropenia [172]
Vorinostat	In vitro study with UGT-overexpressing HEK-293 cell homogenates, human liver microsomes and human colon homogenates [164]	UGT1A7*2	No detectable differences in activity compared with UGT1A7*1a [164]
		UGT1A7*3	No detectable differences in activity compared with UGT1A7*1a [164]
		UGT1A7*4	No detectable differences in activity compared with UGT1A7*1a [164]
		UGT1A8*2	Threefold decrease in activity compared with UGT1A8*1a [164]
		UGT1A8*3	No detectable activity [164]
		UGT1A10*2a	No detectable activity [164]
		UGT2B17*2	Reduced activity, gene expression and enzyme affinity [164]
TAS-103	Phase I clinical trial of weekly TAS-103 in patients with advanced cancer [175]	UGT1A1*28	No correlation with TAS-103, TAS-103 glucuronide or clearance although it may correlate with severe neutropenia at high dose levels [175]

^†UGT2B7 haplotype 4 is defined by −45597G; −6682_−6683A; 372A; IVS1+9_IVS1+10A; IVS1+829T; IVS1+985G; IVS1+999C; IVS1+1250G; 801T; IVS4+185C. FEC100: 5-fluorouracil 500 mg/m², epirubicin 100 mg/m² and cyclophosphamide 500 mg/m²; UGT: Uridine 5′-diphosphoglucuronosyltransferase.

Raloxifene

Raloxifene is an oral selective estrogen receptor modulator (reviewed in [130]). It has been approved by the FDA as adjuvant therapy for reducing the risk for invasive breast cancer in postmenopausal women with either osteoporosis or at increased risk for invasive breast cancer [131]. Raloxifene appears to be as effective as tamoxifen in reducing invasive breast cancer risk and has a lower risk of adverse events [132]. The drug is also used for the prevention and treatment of osteoporosis in postmenopausal women (reviewed in [133]).

Raloxifene undergoes extensive presystemic glucuronidation and enterohepatic circulation [130,133,134]. The metabolites produced are raloxifene-4′-β-glucuronide and raloxifene-6-β-glucuronide, both of which have low affinity for the estrogen receptor and are less potent at inhibiting cell proliferation than raloxifene [135]. The main metabolite found in human plasma and in jejunal and ileal microsomes is raloxifene-4′-β-glucuronide [134,136,137], which is formed mainly by UGT1A10 followed by UGT1A8. Hepatic UGT1A9, UGT1A1, UGT1A3 and UGT2B7 also contribute to the reaction to a lower extent [136,138]. Human liver microsomes, on the other hand, favor glucuronidation at the 6-position [136]. The main enzyme producing raloxifene-6-β-glucuronide is UGT1A1, but there is also some contribution by extrahepatic UGT1A8 and hepatic UGT1A9 and UGT1A3 [136,138].

Only one study has evaluated the influence of polymorphic variation in UGT, in particular UGT1A1*28, on raloxifene glucuronidation. This in vivo study of 57 postmenopausal women treated for osteoporosis measured the concentrations of raloxifene and its two glucuronides in serum samples, and the change in bone mineral density after 1 year of raloxifene therapy [139]. Patients with UGT1A1*28/*28 genotype had increased exposure to raloxifene and its glucuronides, and a significant increase in hip bone mineral density compared with UGT1A1*1/*28 and UGT1A1*1/*1 patients. A significant increase in glucuronides in UGT1A1*28/*28 carriers was unexpected, and the authors hypothesized that it may be due to impaired excretion. Additional studies are needed to clarify the relationship between glucuronidation and UGT1A1*28, and to examine whether the variants have an effect on breast cancer reduction in women undergoing raloxifene treatment.

Epirubicin

Epirubicin, the 4′-epi-isomer of doxorubicin [140], is used for the treatment of advanced breast cancer (reviewed in [141,142]) and gynecological cancers [143,144]. Its primary cellular target is topoisomerase II [140], and one of its main toxic effects is myelosuppresssion [145]. Epirubicin is metabolized quickly in the human body to form epirubicin-glucuronide, epirubicinol and epirubicinol-glucuronide [146,147]. Studies with human liver microsomes expressing specific human UGTs and recombinant enzymes have demonstrated that epirubicin glucuronide is formed by UGT2B7 [148]. No differences in glucuronidation were observed in HEK-293 cells expressing UGT2B7*1a (reference sequence) and the common UGT2B7*2a variant [148]. Subsequent studies in Caucasian human liver microsomes showed that samples containing UGT2B7 haplotype 4 (−45597G; −6682_−6683A; 372A; IVS1+9_IVS1+10A; IVS1+829T; IVS1+985G; IVS1+999C; IVS1+1250G; 801T; IVS4+185C) had a statistically significant 27% average increase in epirubicin glucuronidation compared with the diplotypes without haplotype 4 [149]. The putative functional variants of haplotype 4 are IVS1 +985A>G, +735A>G and +1062C>T. One of these variants, 735A>G (UGT2B7*1c), has recently been associated with higher in vivo and in vitro zidovudine clearance and hepatic UGT2B7 expression [150]. The effect of this allele and the other two putative functional variants in the in vivo glucuronidation of epirubicin need to be tested in clinical studies.

The UGT2B7 promoter variant −161C/T has been associated with differences in morphine glucuronidation in acute pain patients. In this study, −161C/T was in complete linkage disequilibrium with UGT2B7*2a, and individuals with −161C/C and 802C/C genotypes had reduced morphine glucuronidation [151]. However, other clinical studies have not found an association between these two variants and either morphine glucuronidation to morphine serum ratios or morphine analgesic effect [152-154]. A recent pharmacogenetic study of epirubicin in breast cancer patients studied the impact of −161C/T on pharmacokinetics and toxicity [155]. Patients with the CC genotype had a significant decrease in epirubicin clearance (88.9 l/h) compared with CT and TT patients (129 l/h). Incidence rates of grade 3–4 leukopenia during cycle 1 were also higher in CC carriers (78%) versus CT and TT (48%) [155]. As the functional significance of −161C/T has not been proven, future studies are needed to validate these findings and assess the impact of haplotype 4 on epirubicin pharmacokinetics and pharmacodynamics.

Vorinostat

Vorinostat, also known as suberoylanilide hydroxamic acid, is an oral inhibitor of histone deacetylases. It was approved in 2006 by the FDA for the treatment of cutaneous T cell lymphoma [156]. It is also being tested for the treatment of a number of solid malignancies [157-159]. Toxicity includes thrombocytopenia, diarrhea, nausea and anorexia [158-160].

Vorinostat undergoes extensive metabolism to two inactive metabolites: an O-glucuronide and 4-anilino-4-oxobutanoic acid [161-163]. In vitro experiments with UGT overexpressing cell homogenates showed high levels of vorinostat glucuronidation metabolism by hepatic UGT2B17 and UGT1A9, and extrahepatic UGT1A8 and UGT1A10 [164]. UGT1A3 and UGT1A7 also glucuronidated vorinostat but to a lower extent. Another study of recombinant UGTs confirmed the major contribution of UGT2B17 to vorinostat glucuronidation and also identified UGT1A1 and UGT2B7 as important hepatic isoforms [165]. In vitro experiments investigated the effect of the copy-number variation in UGT2B17 (UGT2B17*2) and missense polymorphisms in UGT1A7, UGT1A8 and UGT1A10 on vorinostat glucuronidation (Table 4) [164]. Human liver microsomes with UGT2B17*2/*2 genotype had significantly lower vorinostat glucuronidation activity and UGT2B17 mRNA levels compared with livers with at least one UGT2B17*1 allele. Similar levels of glucuronidation and gene expression were observed for individuals with one or two UGT2B17*1 alleles. Homozygotes for UGT2B17*2 also had lower affinity (higher Michaelis constant value) for vorinostat compared with homozygotes for UGT2B17*1. Regarding the polymorphisms in extra hepatic tissues, UGT1A7*2, *3 and *4 did not have an impact on vorinostat glucuronidation, while UGT1A8*2 had decreased glucuronidation capacity, and both UGT1A8*3 and UGT1A10*2a exhibited no glucuronidation activity. Clinical studies are needed to further explore the association of UGT polymorphic variation and response to vorinostat.

Flavopiridol

Flavopiridol is a cyclin-dependent kinase inhibitor in clinical development for the treatment of chronic lymphocytic leukemia [166]. It has a dose-limiting toxicity of secretory diarrhea when administered as a 72-h continuous infusion [167-169] attributable to luminal exposure to flavopiridol [167].

Studies in human liver microsomes have shown that flavopiridol forms two glucuronides. The major metabolite is 7-O-β-glucopyranuronosyl-flavopiridol, formed mainly by UGT1A9 and to a minor extent by UGT1A1, UGT1A4, UGT1A8 and UGT1A10 [170,171]. 5-O-β-glucopyranuronosyl-flavopiridol, the minor glucuronide, is formed by UGT1A1, UGT1A4 and to a lower extent by UGT1A9 [170,171]. In a study with Caucasian human livers, no association was observed between flavopiridol glucuronidation, UGT1A9 mRNA levels and UGT1A9*1b and I399C>T polymorphisms [171]. In vitro hepatic flavopiridol glucuronidation also does not appear to be affected by the rare coding variants, UGT1A9*2 (C3Y) and UGT1A9*3 (M33T) [80].

Flavopiridol glucuronidation has correlated with toxicity. A clinical trial of 22 metastatic renal cancer patients used the ratio of flavopiridol glucuronide to parent drug in plasma (metabolic ratios) at the end of the infusion during cycle 1 as a marker for UGT activity [167]. The metabolic ratios showed a bimodal distribution. Patients who experienced diarrhea had significantly lower glucuronidation activity than those without toxicity. Correlation between glucuronidation and response rates could not be examined, as the drug was inactive in this patient cohort.

A pharmacogenetic study administered flavopiridol as an intravenous infusion for 1 h to 55 patients with refractory neoplasms [172]. The dose-limiting toxicities observed were neutropenia, fatigue and diarrhea. The study investigated the effect of UGT1A1*28 and found no association between the variant allele and either pharmacokinetics or the occurrence and severity of diarrhea and neutropenia. This is consistent with in vitro evidence showing that UGT1A1 plays a minor role in flavopiridol glucuronidation.

TAS-103

TAS-103 inhibits topoisomerases I and II and the function of the signal-recognition particle in directing the delivery of secretory proteins [173]. Correlation studies with human liver microsomes showed that TAS-103 is glucuronidated mainly by UGT1A1 [174]. A clinical trial of TAS-103 administered weekly at various doses investigated the influence of UGT1A1*28 on pharmacokinetics and severe toxicity [175]. Drug-related toxicities included neutropenia and mild thrombocytopenia. There was no relationship between TAS-103 and TAS-103G concentrations and UGT1A1*28. At 130- and 160-mg/m² doses, the majority of the UGT1A1*28/*28 carriers experienced dose limiting toxicities (67%) compared with 40% of the individuals with UGT1A1*28/*1 and 11% of those with UGT1A1*1/*1. The number of patients in each dose group was small, however, and further studies are needed to evaluate the effect of UGT1A1 genetic variation on response to TAS-103.

Additional genetic variation to be investigated in future studies

This article focused on polymorphisms associated with response to anticancer agents. Genetic variation also exists in other UGT genes and could be investigated in future studies of drugs metabolized by these specific UGT isoforms. Among these additional candidate polymorphisms is UGT2B15*2 (235G>T, D85Y) [176], a common allele (reviewed in [9]) that is associated with reduced in vitro and in vivo glucuronidation of oxazepam [177-179] and diminished in vivo clearance of lorazepam [180]. Another important and common polymorphism is UGT1A6*2 (S7A/T181A/R184S) [181], which appears to play an important role in the glucuronidation of UGT1A6 substrates, although these effects are substrate dependent in vitro [181-183]. In vivo, UGT1A6*2 has been associated with reduced exposure to acetaminophen and salicylic acid [184,185]. Other UGT1A6 variants, present at a lower frequency (1–2%), include UGT1A6*3 (R184S) and UGT1A6*4 (T181A), but their functionality has not been studied (reviewed in [9]). Additional UGT1A9 polymorphisms to those discussed include UGT1A9*3 (M33T), which also shows a substrate-dependent effect in vitro and is present at an allele frequency of 4% in Caucasians [80,186].

Discussion

Many anticancer agents are metabolized by glucuronidation. Polymorphisms in the UGT genes may affect how patients respond to these drugs. Irinotecan, a drug used for treating metatastic colorectal cancer and other tumor types, has been studied in multiple pharmacogenetic studies aiming to determine how genetic variation in UGT1A influences drug-treatment response. Many studies have demonstrated how UGT1A1*28 and UGT1A1*6 increase the risk for experiencing severe neutropenia [32,34-38,41,42,46,48,49,59,60,64,88], the most frequent dose-limiting toxicity of irinotecan treatment. Although some studies suggest genetic variation in UGT1A9 and UGT1A7 may also influence response to irinotecan treatment [46,47,48,49,60,70], no firm conclusion can be drawn as there is high linkage between allelic variation in these genes and both UGT1A1*28 and UGT1A1*6 [42,47,48,70,79-82,86]. The latest studies are using haplotype approaches to study the combined effects of UGT polymorphisms on irinotecan pharmacodynamics and pharmacokinetics [42,62,70,82]. Genotype-directed studies are also being conducted to individualize treatment and find optimal irinotecan doses in patients, depending on their UGT1A1 genotypes [92-95]. Tamoxifen, used in the treatment of estrogen-receptor-positive breast cancer, is metabolized extensively in the human body. In vitro studies have investigated the effect of several UGT variants on glucuronidation of the drug and its active metabolites. Candidate polymorphisms to be studied in vivo for their capacity to predict clinical response include UGT2B7*2a, UGT1A8*3 and UGT1A4*3b [119,129]. Epirubicin, used to treat advanced breast and gynecological cancers, causes leukopenia. Incidence of severe toxicity in breast cancer patients taking epirubicin has been associated with the UGT2B7 −161C/T variant [155]. Future studies are needed to validate these findings and study the effect of UGT2B7 haplotype 4 on drug response [149]. In vitro studies of vorinostat, used in the treatment of cutaneous T cell lymphoma and solid malignancies, suggest that UGT2B17*2 and UGT1A8*2 may influence how individuals react to the drug [164]. In conclusion, candidate polymorphisms in the UGT genes have been identified for a number of anticancer drugs. Future studies are needed to validate findings and clinically test whether genotyping will help predict response to these drugs.

Conclusion

Genetic variation in the UGT gene is associated with response to anticancer agents. The best example to demonstrate the clinical impact of UGT polymorphisms on anticancer treatment is irinotecan, used in colorectal cancer therapy. The reduced-function alleles UGT1A1*28 and UGT1A1*6 predict for the risk of severe irinotecan toxicity and help identify cancer patients who could benefit from reduced doses of irinotecan or alternative chemotherapy. Functional variants in the UGT1A4, UGT1A8, UGT2B7 and UGT2B17 genes also predict the in vitro or in vivo glucuronidation of tamoxifen, epirubicin and vorinostat, although the functional consequences of these variants still need to be tested in the clinical setting.

Future perspective

The results from studies that have been conducted thus far are promising, and future studies will continue to identify and investigate the functional consequences of individual polymorphic variants and haplotypes to predict drug pharmacokinetics and pharmacodynamics.

Prospective studies of the effect of genetic variation in candidate genes should be conducted, ensuring that the study design has adequate statistical power and controls to account for the effects of covariates. Genetic testing will be more widely used prior to treatment in order to optimize dosage, predict risk for adverse effects and help identify patients who could benefit from either reduced doses or other chemotherapy agents.

Although selecting genes associated with drug disposition has shown promising results for irinotecan patients, new technological advances will allow researchers to adopt genome-wide approaches involving single nucleotide polymorphisms arrays, microarrays and proteomics to select new candidate genes and analyze extended haplotypes.