Introduction

Irinotecan is a key drug used to treat cervical and ovarian cancers; however, ∼20–35% of patients treated with irinotecan experience severe diarrhea and neutropenia.1, 2, 3, 4, 5 As a prodrug, irinotecan is activated by hydrolysis to SN-38, a potent topoisomerase I inhibitor, and inactivated through conversion to SN-38 glucuronide (SN-38G) by UDP-glucuronosyltransferase 1A1 (UGT1A1).6 Gene transcription efficiency is inversely correlated with the UGT1A1 gene polymorphism UGT1A1*28.7 Moreover, as UGT1A1 catalyzes bilirubin glucuronidation, homozygosity for the TA7 allele is associated with Gilbert’s syndrome, a common mild hyperbilirubinemia.8, 9 Three variants (UGT1A1*6, UGT1A1*27 and UGT1A1*28) are associated with reduced enzyme activity and increased irinotecan toxicity,10 and the frequencies of these variants differ between Asian and Caucasian populations.10, 11, 12, 13 These polymorphisms directly reduce UGT enzyme activity.14 UGT1A1*6 also can cause a decrease in metabolic activity and is not linked with UGT1A1*28.15, 16, 17

The U.S. Food and Drug Administration recommends that a dose reduction of irinotecan should be considered in patients homozygous for UGT1A1*28. Furthermore, because of reported ethnic differences, homozygous UGT1A1*6/*6 and UGT1A1*28/*28 genotypes and the heterozygous UGT1A1*6/*28 genotype are seen as poor metabolizers (PMs) of irinotecan. Indeed, patients with UGT1A1 deficiencies experience irinotecan-dependent neutropenia and diarrhea in a dose-dependent manner;18 however, contradictory results have been reported for the association between UGT1A1 genotypes and toxicity profiles in patients receiving low-dose irinotecan.19, 20, 21 Moreover, while some studies have suggested its association with severe neutropenia, total bilirubin is yet to be validated as a marker for predicting the risk of severe neutropenia in gynecological cancer patients treated with irinotecan in the absence of UGT1A1 genetic information.22, 23, 24, 25

To date, only a few studies have described individualized chemotherapy treatment based on the assessment of genetic polymorphisms in gynecologic cancer or on the administration of lower doses of irinotecan for the treatment of gynecologic cancer compared with the doses used for other solid tumors.19 Here, we retrospectively investigated the associations of UGT1A1 polymorphisms with serum bilirubin levels and severe adverse effects in Japanese patients who received irinotecan for the treatment of gynecologic cancer.

Patients and methods

Patients

Blood samples were collected from 415 Japanese patients who visited the menopausal clinic or cancer follow-up clinic of the Department of Obstetrics and Gynecology, Keio University Hospital (Tokyo, Japan). Fifty-three patients who had cervical or ovarian cancer and received irinotecan-containing chemotherapy were retrospectively studied. An additional 362 patients who did not receive irinotecan therapy were chosen at random. Patients with pathological hyperbilirubinemia, including obstructive jaundice, and non-Japanese patients were excluded from the study. Informed consent was obtained from all patients before using their blood. The study was conducted with the approval of the ethics committee of the School of Medicine, Keio University (approval numbers: 20050135, 20060091 and 20070081).

UGT1A1 genotyping and classification

Genomic DNA was extracted by standard procedures. UGT1A1*6 (211G→A), UGT1A1*27 (686C→A) and UGT1A1*28 [(TA)6TAA→(TA)7TAA] polymorphisms were genotyped using an Invader UGT1A1 Molecular Assay kit (Third Wave Technologies, Madison, WI, USA) as previously described.26 UGT1A1 polymorphisms were classified into two groups, as recommended by the Pharmaceuticals and Medical Devices Agency, Japan (PMDA): the PM group and the extensive metabolizer (EM) group. The PM group included homozygous UGT1A1*6/*6 and UGT1A1*28/*28 genotypes and the heterozygous UGT1A1*6/*28 genotype. Patients who were not classified as PMs were assigned to the EM group.

Schedule of irinotecan treatment and evaluation of adverse events

The treatment regimen consisted of either 60 mg m−2 irinotecan on days 1, 8 and 15 and 60 mg m−2 cisplatin on day 1 every 4 weeks or 100 mg m−2 irinotecan alone on days 1, 8 and 15 every 4 weeks. Physical examinations and blood tests were retrospectively reviewed to evaluate toxicity and adverse events in irinotecan-treated patients, and their grades were assessed prospectively according to the common terminology criteria for adverse events, version 3.0.27 The monitoring period for adverse events and the worst grade after irinotecan therapy were evaluated until 6 weeks after last irinotecan administration.

Association of irinotecan toxicity with serum bilirubin and UGT1A1 genotypes

Serum total bilirubin was measured for 392 patients, including 49 patients treated with irinotecan. The highest total bilirubin value was adopted from the patient’s own personal historical data in the hospital. Bilirubin measurement was conducted prior to surgery or any chemotherapy in patients given irinotecan, because the bilirubin level may increase after irinotecan therapy. For patients in whom the total bilirubin levels were 1.4 mg dl−1 or greater, hyperbilirubinemia was diagnosed according to the normal range in the hospital. The relationships between serum bilirubin and UGT1A1 genotypes and between serum bilirubin and irinotecan toxicity were statistically analyzed.

Statistical analyses

Statistical analysis was performed with Exel2011, Microsoft Office: mac2011 (Microsoft Japan Co., Ltd., Tokyo, Japan) with the add-in software Statcel 3 (OMS, Saitama, Japan) and Prism 6 (GraphPad Software, Inc., La Jolla, CA, USA) using the indicated tests. The F-test was used to verify the heterogeneity of variances. Nonparametrically distributed variables were assessed using the Mann–Whitney U-test. The chi-square test or Fisher’s exact test was performed for contingency tables. Results with P-values of less than 0.05 were considered significant.

Results

Patient characteristics

Blood samples from a total of 415 patients, including 11 cervical cancer patients, 107 ovarian cancer patients, 225 endometrial cancer patients, 26 ovarian and endometrial double cancer patients, 8 myoma or endometriotic cyst and 38 menopausal disorder were analyzed for UGT1A1 polymorphisms (Table 1). Among these patients, 53 were treated with irinotecan.

Table 1 Characteristics of the patients included in this study
Full size table

UGT1A1 genotype and allele frequency

Genotype and allele frequencies of UGT1A1*6, UGT1A1*27 and UGT1A1*28 in 415 patients, including 53 patients that received irinotecan and 362 without irinotecan treatment, are summarized in Supplementary Table 1.

Two patients (3.8%) treated with irinotecan were homozygous for UGT1A1*6, and none were homozygous for UGT1A1*28. Fourteen patients (26.4%) were heterozygous for UGT1A1*6, and 9 patients (17.0%) were heterozygous for UGT1A1*28. One patient was simultaneously heterozygous for UGT1A1*6 and UGT1A1*28 (data not shown). No irinotecan-treated patients were heterozygous or homozygous for UGT1A1*27.

None were homozygous for UGT1A1*27, whereas 17 (4.1%) and 7 (1.9%) patients without irinotecan treatment were homozygous for UGT1A1*6 and UGT1A1*28, respectively.

Of all cases, all UGT1A1*27 carriers also harbored UGT1A1*28 alleles (data not shown). Moreover, 14 patients (9.2%) were simultaneously heterozygous for UGT1A1*6 and UGT1A1*28 (data not shown).

UGT1A1*6 and UGT1A1*28 groups did not differ significantly between patients receiving or not receiving irinotecan (P=0.802 and 0.464, respectively).

The PM and EM groups did not differ significantly between the total patient population and irinotecan-treated patients (P=0.605; Supplementary Table 2).

Correlation between the UGT1A1 genotype and adverse effects of irinotecan

Genotypes and adverse event assessments for 53 patients (11 cervical cancer, 40 ovarian cancer and 2 ovarian and endometrial double cancer patients) who received irinotecan are summarized in Table 2.27 Thirty-five patients had grades 0–2 absolute neutrophil count, whereas 18 patients had grades 3–4 absolute neutrophil count (Table 2). In addition, 48 patients had grades 0–2 diarrhea, and 5 patients had grades 3–4 diarrhea (Table 2). The occurrence of grades 3–4 absolute neutrophil count and diarrhea was significantly more frequent in PM-group patients than in EM-group patients (odds ratio: 16.03, P=0.0348 and odds ratio: 31.33, P=0.0209, respectively). However, UGT1A1*6 or UGT1A1*28 alone was not associated with absolute neutrophil count.

Table 2 Adverse events and effects of the UGT1A1 polymorphism in patients administered irinotecan
Full size table

Three cases in Table 3 were classified as PMs of irinotecan (the PM group). Two of these cases were homozygous for UGT1A1*6, and one was simultaneously heterozygous for UGT1A1*6 and UGT1A1*28. All three had a grade 3 or 4 absolute neutrophil count, and two had grade 3 diarrhea.

Table 3 Characteristics of three patients classified as PM, who were given irinotecan
Full size table

Relationship between serum bilirubin and UGT1A1 genotypes or irinotecan toxicity

Total bilirubin levels were significantly higher in the PM group (n=38) than in the EM group (n=354; P<0.0001; Figure 1). Sixty-nine out of 392 patients had been diagnosed with hyperbilirubinemia (⩾1.4 mg dl−1) more than once in their history (Table 4). The occurrence of hyperbilirubinemia was significantly increased in heterozygous or homozygous UGT1A1*28 cases compared with heterozygous or homozygous UGT1A1*6 cases. Furthermore, the occurrence of hyperbilirubinemia was significantly increased in PM-group cases compared with EM-group cases (odds ratio: 6.08, P<0.0001; Table 4). However, serum bilirubin levels were not associated with irinotecan toxicity-induced grade 3 or 4 absolute neutrophil count or diarrhea in the 49 patients who were treated with irinotecan (P>0.05 for both; Table 5).

Figure 1
figure 1

Association between UGT1A1 genotypes and total bilirubin levels. Serum total bilirubin levels were significantly higher in the PM group (n=38) compared with the EM group (n=354), as determined by analysis of 392 patients who had data on serum bilirubin levels (P<0.0001).

Full size image
Table 4 UGT1A1 genotypes and hyperbilirubinemia
Full size table
Table 5 Associations between serum bilirubin levels and irinotecan-related adverse events
Full size table

Discussion

Previous studies have revealed ethnic differences in UGT1A1 haplotypes among Caucasians, African-Americans and Asians.7, 12, 13, 14 Conversely, UGT1A1*6 and UGT1A1*27 polymorphisms have been identified only in Asians.8, 13 Our results for the allele frequencies of UGT1A1*6, UGT1A1*27 and UGT1A1*28 are similar to those of previous reports.

A prior study demonstrates that the UGT1A1*6 genotype is related to lower SN-38 glucuronidation and a higher frequency of grade 3/4 toxicities in patients receiving irinotecan and cisplatin.28 Moreover, a grade 3/4 absolute neutrophil count is significantly more frequent in patients who have the UGT1A1*6 genotype and are undergoing irinotecan-based chemotherapy, and homozygosity for the UGT1A1*6 allele is associated with altered SN-38 disposition, potentially increasing the risk of severe neutropenia in Asian cancer patients.29 Our data could not support this theory, most likely because we did not have enough UGT1A1*6 and UGT1A1*28 cases in this study.

Since UGT1A1*27 is very rare, its clinical impact has not yet been fully elucidated. The UGT1A1*27 and UGT1A1*28 alleles are linked; therefore, UGT1A1*28 genotyping is thought to cover UGT1A1*27 carriers.30 In our study, all four UGT1A1*27 carriers also harbored UGT1A1*28 alleles. Some studies have yielded contradictory results for the relationship between UGT1A1 genotypes and toxicity profiles in patients receiving low-dose irinotecan.19, 20 For example, the risk of hematologic toxicity is strongly associated with the UGT1A1*28 genotype at higher doses (>150 mg m−2), but not at lower doses (⩽150 mg m−2).20 However, a recent meta-analysis revealed that the UGT1A1*28/*28 genotype was associated with an increased risk of neutropenia at all doses of irinotecan.21

Irinotecan is often administered as combination therapy with cisplatin for gynecologic cancer. In a previous study of 177 Japanese cancer patients treated with irinotecan as a single agent or as combination chemotherapy with cisplatin,31 significant associations between UGT1A1*6 or UGT1A1*28 and neutropenia were observed for patients in both groups. Moreover, significant associations between UGT1A1*6 or UGT1A1*28 and diarrhea were also observed for patients who received irinotecan as a single therapy, but not in patients who received combination therapy with cisplatin.31 Of the three PM patients who received irinotecan in our study, one patient was administered irinotecan as a single agent and two patients were given combination chemotherapy with cisplatin (Table 3). Thus, our results demonstrated that significant associations between UGT1A1*6 or UGT1A1*28 and low absolute neutrophil count or diarrhea occurred in Japanese ovarian cancer patients who received irinotecan as single agent or in combination with cisplatin. These findings are in agreement with those of Takano et al., who first reported that grade 3/4 neutropenia, thrombocytopenia and diarrhea were significantly more frequent in UGT1A1*6 and/or UGT1A1*28 genotypes in Japanese gynecologic cancer patients treated with the irinotecan–cisplatin combination.19 In contrast, our findings contradicted a previous report showing no significant contribution of UGT1A1 genotypes to toxicity in patients treated with low-dose irinotecan monotherapy.20 However, interpatient differences in SN-38G formation, which have been correlated with the occurrence of severe diarrhea, and the lower doses of irinotecan used in the present study may, at least in part, explain these conflicting results.32, 33

Hyperbilirubinemia may be a potential alternative predictive marker for irinotecan toxicity. Indeed, modest elevations in bilirubin (1.0–1.5 mg dl−1) are associated with increased grades 3–4 neutropenia in metastatic colorectal cancer patients treated with weekly irinotecan.22 However, baseline serum bilirubin does not reliably predict overall irinotecan-related toxicity or efficacy in these cases. Interestingly, UGT1A1 polymorphisms are associated with serum bilirubin levels, and UGT1A1*28 has been reported to be associated with Gilbert’s syndrome.8, 9 UGT1A1*6, predominant in east Asia, also leads to lower UGT1A1 expression and hyperbilirubinemia.15, 16, 17 In the current study, hyperbilirubinemia was significantly more frequent in heterozygous or homozygous UGT1A1*28 cases or PM cases (Table 4), but was not associated with irinotecan toxicity in our study (Table 5). Thus, our findings strongly suggested that UGT1A1 genotyping may be a more powerful predictive biomarker than serum bilirubin.

Our study has some limitations. Because of the sampling procedure used, the patients examined here were cancer survivors, and cases of aggressive disease were not included. In addition, UGT1A1 polymorphisms are thought to influence the development of some cancers, including colorectal, breast and ovarian cancers.34 Moreover, an individual’s response to irinotecan is unlikely to be affected by UGT1A1*28 status.35 We also did not distinguish between patients receiving irinotecan+cisplatin versus irinotecan alone; this may have affected the outcome of the study. The range of irinotecan dosages in clinical use is very wide and was determined according to organ-based therapies tested in large-scale clinical studies.21 Moreover, the metabolism of irinotecan to SN-38 and other metabolites varies widely between patients. Therefore, a wide range of doses is required, and dose modifications or variations should be based on genotyping. Further prospective studies on the pharmacokinetics, pharmacodynamics and pharmacogenetics of irinotecan are needed to clarify these findings, especially in patients receiving low-dose irinotecan, in which few clinical trials have been conducted.

In conclusion, homozygosity for UGT1A1*6/*6 and heterozygosity for both UGT1A1*6/*28 were predictive of low absolute neutrophil count and diarrhea induced by irinotecan in Japanese gynecologic cancer patients, whereas serum bilirubin levels were not.