• Open Access

Genotype-directed, dose-finding study of irinotecan in cancer patients with UGT1A1*28 and/or UGT1A1*6 polymorphisms

Authors


16To whom correspondence should be addressed.
E-mail: taroh@cfs.med.osaka-u.ac.jp

Abstract

Irinotecan-induced severe neutropenia is associated with homozygosity for the UGT1A1*28 or UGT1A1*6 alleles. In this study, we determined the maximum-tolerated dose (MTD) of irinotecan in patients with UGT1A1 polymorphisms. Patients who had received chemotherapy other than irinotecan for metastatic gastrointestinal cancer were enrolled. Patients were divided into three groups according to UGT1A1 genotypes: wild-type (*1/*1); heterozygous (*28/*1, *6/*1); or homozygous (*28/*28, *6/*6, *28/*6). Irinotecan was given every 2 weeks for two cycles. The wild-type group received a fixed dose of irinotecan (150 mg/m2) to serve as a reference. The MTD was guided from 75 to 150 mg/m2 by the continual reassessment method in the heterozygous and homozygous groups. Dose-limiting toxicity (DLT) and pharmacokinetics were evaluated during cycle 1. Of 82 patients enrolled, DLT was assessable in 79 patients (wild-type, 40; heterozygous, 20; and homozygous, 19). Dose-limiting toxicity occurred in one patient in the wild-type group, none in the heterozygous group, and six patients (grade 4 neutropenia) in the homozygous group. In the homozygous group, the MTD was 150 mg/m2 and the probability of DLT was 37.4%. The second cycle was delayed because of neutropenia in 56.3% of the patients given the MTD. The AUC0–24 h of SN-38 was significantly greater (< 0.001) and more widely distributed in the homozygous group. Patients homozygous for the UGT1A1*28 or UGT1A1*6 allele can receive irinotecan in a starting dose of 150 mg/m2, but many required dose reductions or delayed treatment in subsequent cycles. UMIN Clinical Trial Registration number: UMIN000000618. (Cancer Sci 2011; 102: 1868–1873)

Irinotecan, a semisynthetic camptothecin derivative with topoisomerase I-inhibiting activity,(1–3) entered clinical studies in the early 1990s and has been shown to be an effective anticancer drug against several malignancies. Irinotecan is a prodrug that is converted to its active metabolite, SN-38, by carboxylesterase. SN-38 is converted to an inactive metabolite, SN-38G, by UGTs. Irinotecan-associated adverse events, such as myelosuppression and diarrhea, are significantly correlated with the AUC of irinotecan, SN-38, and SN-38G.(1–3)

One of the isoforms of UGT, UGT1A1, is the main enzyme involved in SN-38 glucuronidation. Several studies have reported correlations between UGT1A1 polymorphisms and irinotecan-associated adverse events,(4–6) and the efficiency of SN-38 glucuronidation is decreased in individuals homozygous for the UGT1A1*28 allele.(7) In 2005, the US Food and Drug Administration revised the package insert for irinotecan (Camptosar, Yakult Honsha Co. Ltd., Tokyo, Japan), recommending that a reduced dose should be used in these individuals.(8) A subsequent meta-analysis suggested that genetic testing might not be necessary unless the dose of irinotecan is ≤150 mg/m2; however, there was insufficient evidence for doses of approximately 150 mg/m2,(9) and the recommendations for dose adjustments remain unclear.

Although UGT1A1*28 is considered an important predictor of irinotecan-related toxicity, ethnic differences have been reported.(10,11) The allele frequency of UGT1A1*28 is lower in Asians than in Caucasians, and grade 3–4 hematologic toxicity is associated with UGT1A1*6 polymorphisms in Asians.(10) In addition, a recent Japanese study revealed that severe adverse events were associated with double heterozygosity (UGT1A1*28/*6).(11) Adverse events are related to the pharmacokinetic properties of the drug, and the AUC ratio of SN-38G to SN-38 (AUCSN-38G/AUCSN-38) was significantly reduced in UGT1A1*28/*6 patients.(12,13) Thus, UGT1A1*6 appears to be another important predictor of irinotecan-induced adverse events.

In this context, this study was designed to determine the MTD of irinotecan in patients with gastrointestinal cancer whose UGT1A1 genotypes were UGT1A1*28/*28, *6/*6, or *28/*6. The CRM(14,15) was used to determine dose escalation and reductions. Compared with the traditional phase I design, the CRM can incorporate the actual drug doses into the analytical model and evaluate the frequency of adverse events more accurately, particularly in small groups of patients, such as those who are homozygous for UGT1A1*28 or UGT1A1*6. Thus, the CRM was considered appropriate for our study objectives. We also investigated the pharmacokinetics and toxicity of irinotecan in patients with the UGT1A1*28 and UGT1A1*6 alleles.

Patients and Methods

Patients.  Patients meeting the following criteria were eligible for the study: histologically confirmed and inoperable gastrointestinal cancer; ≥20 years old; Eastern Cooperative Oncology Group performance status of 0–1; no prior treatment with irinotecan; a wash-out period of 21 days after previous chemotherapy; adequate bone marrow function (leukocyte count 3000–12 000/mm3; hemoglobin concentration ≥8.5 g/dL; platelet count ≥ 100 000/mm3); serum creatinine ≤ 1.5 mg/dL; total bilirubin ≤2.0 mg/dL; and aspartate aminotransferase and alanine aminotransferase ≤100 IU/L. Written informed consent was obtained from all participants. Patients were excluded if they had severe ascites or had received chest or abdominal radiotherapy. The study protocol was approved by the Institutional Review Board at each institution. An independent Data and Safety Monitoring Committee evaluated safety throughout the study.

UGT1A1 genotyping assay.  Genomic DNA was extracted from peripheral blood using a QIAamp blood kit (Qiagen, Hilden, Germany). An Invader UGT1A1 Molecular Assay kit (Third Wave Technologies, Madison, WI, USA) was used to genotype the UGT1A1*28 and UGT1A1*6 polymorphisms.

Classification of UGT1A1 polymorphisms.  We classified the UGT1A1 polymorphisms into three groups: wild-type (*1/*1), heterozygous (*28/*1, *6/*1), and homozygous (*28/*28, *6/*6, *28/*6). The double heterozygous state (*28/*6) was included within the homozygous group, taking into account the findings of previous studies.(11,13) A recent study found no evidence of alleles *28 and *6 existing on the same chromosome; patients harboring *28 and *6 on the same chromosome either do not exist or are extremely rare.(16) Therefore, concurrent *28 and *6 homozygosity was considered irrelevant.

Treatment schedule.  Irinotecan was given i.v. over the course of 90 min of every 14-day cycle, for only two cycles. The wild-type group received a fixed dose of 150 mg/m2 as a reference. This is a borderline dose between the low and medium dose levels, as proposed by Hoskins et al.,(9) and is the upper limit of the approved dose of irinotecan in Japan. The starting dose was 100 mg/m2 in the heterozygous group and 75 mg/m2 in the homozygous group. The dose was escalated in increments of 25 mg/m2 up to 150 mg/m2, as described below. The study treatment comprised two cycles, unless unacceptable toxicity developed during the first cycle, or the patient withdrew consent.

Safety was evaluated in the first and second cycles, and DLT was only assessed in the first cycle. Objective clinical evaluations, blood counts, and hepatic and renal function tests were carried out on days 1 and 8 of each cycle. Dose-limiting toxicity was defined as grade 4 neutropenia, grade 4 thrombocytopenia, febrile neutropenia (neutrophil count < 1000/mm3 and fever ≥ 38.5°C), or grade 3 diarrhea. If DLT occurred in the first cycle, the dose was reduced by one dose level in the second cycle. Toxicity was evaluated according to the Common Terminology Criteria for Adverse Events version 3.0.

Pharmacokinetic assay.  Venous blood for pharmacokinetic analysis was collected in sodium-heparinized, evacuated tubes on days 1 and 2 of the first cycle, before infusion of irinotecan, at the end of infusion, and at 1, 2, 4, 7, and 24 h after infusion. The plasma concentrations of intact irinotecan, SN-38, and SN-38G were determined by HPLC, as previously described.(17) The AUC0–24 h was calculated using WinNonlin software version 5.0.1 (Pharsight, Mountain View, CA, USA).

Dose escalation/reductions: design and statistical considerations.  Eligible patients underwent genotyping and were assigned to the wild-type group, heterozygous group, or homozygous group. They were then registered at the data center. All patients who received at least one dose of irinotecan without major protocol violations were included in the safety and pharmacokinetic analyses.

In the wild-type group, we estimated the probability of DLT occurring at a dose of 150 mg/m2. A sample size of 40 was planned, assuming that the probability of DLT would be 10% (maximum 20%) with 95% confidence limits.

In the heterozygous and homozygous groups, dose escalation and reductions were carried out according to the CRM. We used a logistic regression model to determine the relationship between dose and toxicity. The model was updated based on the patients’ responses, using a Bayesian approach. After enrolling the first patient, the doses given to subsequent patients were determined by the CRM. Each subsequent patient was treated at the dose level where the probability of DLT was closest to 30%. Dose-limiting toxicity was assessed in a maximum of three patients at the same time and dose. The dose was increased or decreased by 25 mg/m2. The MTD was defined as the dose level at which nearest to 30% of patients were expected to have DLT. The recommended dose was determined based on the results obtained during the first two cycles. Simulation studies indicated that 10–20 patients were required to estimate the MTD. The decision to continue or stop the study was made after safety evaluation of the first 10 patients.

Because of a treatment-related death, patient enrolment was temporarily halted and the protocol was revised. The dose of irinotecan in the homozygous group was reduced by two levels in the second cycle if the patient had grade 3–4 neutropenia in the first cycle.

In accordance with the advice of the Data and Safety Monitoring Committee, medical experts and biostatisticians, all eligible patients in the homozygous group were included in the analysis of MTD, DLT, and the toxicity data, irrespective of protocol amendments.

The Cochran–Armitage trend test was used to analyze trends in grade 3–4 adverse events across the different genotypes. Fisher’s exact test was used to compare the frequency of toxicity among the wild-type, heterozygous, and homozygous groups. Pearson’s correlation coefficient was used to assess the relationships between laboratory test data and pharmacokinetic parameters during the first cycle. The association between pharmacokinetic parameters and genotype was evaluated using the Cochran–Armitage trend test. Levene’s test was used to assess the homogeneity of variances in SN-38 among the genotypes. All analyses were carried out using SAS software version 8.2 (SAS Institute, Cary, NC, USA).

Results

Patient characteristics.  Between November 2006 and October 2008, 82 patients were enrolled at 12 institutions and assigned to the wild-type (n = 41), heterozygous (n = 20; *28/*1 [n = 8], *6/*1 [n = 12]), or homozygous (n = 21; *28/*28 [n = 3], *6/*6 [n = 12], *28/*6 [n = 6]) groups. Toxicity and pharmacokinetic parameters were evaluated in 81 patients, excluding one patient in the homozygous group who withdrew consent before treatment. After the first dose, two patients were deemed ineligible. Therefore, 79 patients were eligible for DLT analysis (Table 1).

Table 1.   Disposition and baseline characteristics of patients with gastrointestinal cancer who participated in this study (n = 82)
CharacteristicsWild-type groupHeterozygous groupHomozygous group
n%n%n%
  1. Patients were divided into groups according to UGT1A1 genotype: wild-type (*1/*1); heterozygous (*28/*1, *6/*1); or homozygous (*28/*28, *6/*6, *28/*6). DLT, dose-limiting toxicity; ECOG, Eastern Cooperative Oncology Group; —, not applicable.

Patients enrolled412021
Consent withdrawn001
Patients receiving study drug412020
Eligible402019
Not eligible101
DLT analysis402019
Toxicity analysis412020
Pharmacokinetic analysis412020
Sex
 Male2253.71050.01365.0
 Female1946.31050.0735.0
Age (years)
 Median626366
 Range21–8847–8038–78
ECOG performance status
 02765.91260.01155.0
 11434.1840.0945.0
Adenocarcinoma histology
 Stomach1946.3945.0420.0
 Colorectal2253.71155.01680.0
Total bilirubin (mg/dL)
 Median0.50.60.8
 Range0.2–1.30.2–1.00.3–2.0
Direct bilirubin (mg/dL)
 Median0.20.20.2
 Range0.0–0.50.1–0.30.1–0.5

Dose escalation and identification of MTD.  The first four patients in each of the heterozygous and homozygous groups showed no DLT, so the dose was increased to 150 mg/m2 according to the CRM (Fig. 1). At 150 mg/m2, DLT occurred in one patient in the wild-type group (grade 3 anorexia and fatigue) and in six patients in the homozygous group (*28/*28 [n = 1], *6/*6 [n = 4], *28/*6 [n = 1]) (grade 4 neutropenia, 6; grade 3 diarrhea, 1), but in no patients in the heterozygous group. Based on these data, the probability of DLT at 150 mg/m2 was 2.5% in the wild-type group (1/40 patients; 95% CI, 0.1–13.2), 5.9% in the heterozygous group (0/16 patients; 80% CI based on the CRM, 2.2–11.2%), and 37.4% in the homozygous group (6/15 patients; 80% CI based on the CRM, 22.8–52.7%). In the homozygous group, the initial dose of irinotecan (150 mg/m2) was determined to be the MTD, whereas the MTD in the heterozygous group was estimated to be >150 mg/m2.

Figure 1.

 Probability of dose-limiting toxicity (DLT) of irinotecan in patients with gastrointestinal cancer, determined using the continual reassessment method. Patients were grouped as heterozygous (A) or homozygous (B) according to their UGT1A1 genotype. (○), Patients registered; (•), occurrence of DLT. CI, confidence interval.

Toxicity.  The major adverse events in patients treated with 150 mg/m2 irinotecan are listed in Table 2. The most frequently observed grade 3–4 toxicities were leukopenia and neutropenia. During the first cycle, hematologic toxicity was significantly associated with genotype (< 0.001). Grade 3–4 neutropenia occurred in 9.8% of patients in the wild-type group, 18.8% of patients in the heterozygous group, and 62.5% of patients in the homozygous group. The frequency of severe neutropenia was significantly higher in the homozygous group than in the wild-type and heterozygous groups (< 0.001). A similar trend was observed during the first two cycles (wild-type group, 22.0%; heterozygous group, 25.0%; homozygous group, 81.3%). Unlike hematologic toxicity, non-hematologic toxicity was not associated with genotype and was generally mild to moderate in severity (Table 2).

Table 2.   Common adverse events (grades 3–4) associated with 150 mg/m2 irinotecan in patients with gastrointestinal cancer
Adverse eventsFirst cycleP*P**First and second cycles
Wild-type group (n = 41)Heterozygous group (n = 16)Homozygous group (n = 16)Wild-type group (n = 41)Heterozygous group (n = 16)Homozygous group (n = 16)
G3/G4%G3/G4%G3/G4%G3/G4%G3/G4%G3/G4%
  1. Patients were divided into groups according to UGT1A1 genotype: wild-type (*1/*1); heterozygous (*28/*1, *6/*1); or homozygous (*28/*28, *6/*6, *28/*6). *Cochrane–Armitage trend test; **Fisher’s exact test, wild-type/heterozygous group versus homozygous group. G, grade; NS, not significant.

Hematologic toxic effects
 Leukopenia1/02.40/00.08/156.3<0.001<0.0011/02.42/012.510/275.0
 Neutropenia4/09.83/018.84/662.5<0.001<0.0019/022.03/125.06/781.3
Non-hematologic toxic effects
 Diarrhea0/00.00/00.01/06.3NSNS0/00.00/00.01/06.3
 Fatigue1/02.40/00.00/00.0NSNS1/02.40/00.00/00.0
 Anorexia1/02.40/00.00/00.0NSNS1/02.41/06.30/00.0

On UGT1A1 diplotype analysis, grade 3–4 neutropenia and leukopenia occurred frequently in patients in the homozygous group (Table 3). Grade 3–4 diarrhea occurred in 1/9 patients (11.1%) with *6/*6. The second cycle was delayed in 5/41 patients (12.2%) in the wild-type group, 4/16 patients (25.0%) in the heterozygous group, and 9/16 patients (56.3%) in the homozygous group. The reasons for delaying treatment were neutropenia in seven patients, infection and stomatitis in one, and diarrhea in one. In the homozygous group treated with 150 mg/m2 irinotecan, the dose for the second cycle was reduced by two dose levels (or to 100 mg/m2) in three patients and by one dose level (to 125 mg/m2) in one patient. One patient who received a reduced dose of 100 mg/m2 irinotecan developed grade 4 neutropenia again in the second cycle. Four of 16 patients (25.0%) in the homozygous group completed two cycles of therapy without needing to delay treatment or reduce the dose.

Table 3.   Association between UGT1A1 genotype in patients with gastrointestinal cancer and the most common grade 3–4 adverse events during the first treatment cycle with 150 mg/m2 irinotecan
Adverse events*6/*1 (n = 9)*28/*1 (n = 7)*6/*6 (n = 9)*28/*28 (n = 3)*28/*6 (n = 4)
G3/G4%G3/G4%G3/G4%G3/G4%G3/G4%
  1. G, grade.

Hematologic toxic effects
 Leukopenia0/00.00/00.05/166.72/066.71/025.0
 Neutropenia1/011.12/028.62/466.72/11000/125.0
Non-hematologic toxic effects
 Diarrhea0/00.00/00.01/011.10/00.00/00.0
 Fatigue0/00.00/00.00/00.00/00.00/00.0
 Anorexia0/00.00/00.00/00.00/00.00/00.0

There was one treatment-related death in the homozygous group, which was caused by septic shock with grade 4 neutropenia after the second cycle of irinotecan at a dose of 150 mg/m2. This patient, who was homozygous for UGT1A1*28, had no DLT in the first cycle, and the second cycle was delayed because of prolonged neutropenia.

Relationships between UGT1A1 polymorphisms and pharmacokinetic profile and toxicity of irinotecan.  The AUC0–24 h of SN-38 was significantly higher in the homozygous group than in the wild-type or heterozygous groups (< 0.001) and interpatient variability was also higher in the former group (Table 4). The AUC0–24 h of SN-38G was significantly higher in the wild-type group than in the heterozygous or homozygous groups (= 0.001). The AUCSN-38G/AUCSN-38 ratio was highest in the wild-type group, intermediate in the heterozygous group, and lowest in the homozygous group (< 0.001).

Table 4.   Relationship between UGT1A1 genotype and pharmacokinetic parameters for 150 mg/m2 irinotecan in patients with gastrointestinal cancer
Pharmacokinetic parameterSN-380–24 h (ng × h/mL)SN-38G0–24 h (ng × h/mL)AUC ratio
  1. Analyses were carried out between wild-type versus heterozygous or homozygous groups. AUC, area under the time–concentration curve; AUC ratio, AUCSN-38G/AUCSN-38; SN-38G, SN-38 glucuronide.

Wild-type group (n = 41)264 ± 1141266.8 ± 667.55.03 ± 2.25
Heterozygous group (n = 16)279.6 ± 152.0820.7 ± 378.73.25 ± 1.32
UGT1A1*6/*1 (n = 9)250.2 ± 70.4723.1 ± 252.83.16 ± 1.49
UGT1A1*28/*1 (n = 7)317.3 ± 219.5946.1 ± 490.63.35 ± 1.16
Homozygous group (n = 16)509.8 ± 261.8849.0 ± 561.91.85 ± 1.13
UGT1A1*28/*6 (n = 4)251.3 ± 89.5557.8 ± 148.82.34 ± 0.82
UGT1A1*6/*6 (n = 9)564.9 ± 223.5673.2 ± 304.11.21 ± 0.36
UGT1A1*28/*28 (n = 3)689.0 ± 327.01764.7 ± 631.43.10 ± 1.82
Cochrane–Armitage trend test<0.0010.001<0.001
Levene’s test<0.0010.3100.013

The AUC0–24 h of SN-38 was slightly higher in patients with the *28/*28 or *6/*6 genotypes than in patients with the *28/*6 genotype (Table 4). The AUCSN-38G/AUCSN-38 ratio was slightly lower in patients with the *6/*6 genotype than in those with the *28/*28 or *28/*6 genotypes.

The relationship between adverse events and pharmacokinetic parameters was also analyzed. The AUC0–24 h of SN-38G was not correlated with hematologic toxicity. In contrast, the AUC0–24 h of SN-38 was correlated with the frequency of grade 3–4 leukopenia and neutropenia (r = 0.49, < 0.001). The AUCSN-38G/AUCSN-38 ratio also correlated with the frequencies of grade 3–4 leukopenia (r = 0.25, = 0.023) and neutropenia (r = 0.308, = 0.005).

Discussion

Irinotecan is generally given at a dose of 150 mg/m2 every 2 weeks in Japanese patients with gastrointestinal cancer. This dose was determined based on the results of clinical trials in the 1990s(18,19) and is the upper limit of the approved dose in Japan. The appropriate dosages of cytotoxic agents have been based on the concept of minimizing the risk of no response within the acceptable toxicity limits. However, genetic information has recently been obtained on the metabolism of CPT-11 and its related toxicities. Hoskins et al.(20) showed that the risk of severe hematologic toxicity is higher among patients with the UGT1A1*28/*28 genotype than among those with the UGT1A1*28/*1 or UGT1A1*1/*1 genotypes at medium doses (150–250 mg/m2) and at higher doses (>250–350 mg/m2), but not at lower doses (100–125 mg/m2). However, the results of a recently reported meta-analysis(21) showed that the UGT1A1*28/*28 genotype was associated with an increased risk of neutropenia not only at medium or high doses of irinotecan, but also at low doses (relative risk [RR], 2.43; 80–145 mg/m2). To verify these previous findings in a prospective manner, we needed to reset the MTD according to these genetic factors, so we used irinotecan at doses of 75–150 mg/m2.

Our genotype-directed dose-finding study using the CRM showed that the principal DLT was neutropenia and the MTD of irinotecan was 150 mg/m2 in Japanese patients carrying the UGT1A1 variant alleles. In the heterozygous group, the MTD was estimated to be >150 mg/m2. Our results also showed that the hematologic toxicity of irinotecan at 150 mg/m2 was significantly more severe, and the AUC0–24 h of SN-38 was significantly higher and more widely distributed in patients with two variant alleles than in those with one or no variant alleles. These findings are consistent with the results of previous studies.(12,13,22,23) As described above, UGT1A1 genetic polymorphism is a factor that clearly affects pharmacokinetics, and individual variation in pharmacokinetics was greater in the homozygous group. The recommended dose could not be defined, because it was considered inappropriate to apply the dose obtained by increasing the number of cases to the general population. Moreover, the present study revealed that the AUC0–24 h of SN-38 in the heterozygous group was similar to that in the wild-type group among patients treated with irinotecan at 150 mg/m2. The UGT1A1 polymorphisms were not related to diarrhea in our study, or in previous studies.(10,22,24) Thus, further studies are needed to determine the predictive factors for diarrhea.

When the UGT1A1*28 and UGT1A1*6 alleles were evaluated separately, the incidence of neutropenia and the AUC0–24 h of SN-38 were similar in patients with a homozygous genotype. Although the frequency of patients with UGT1A1*28/*28 has been reported to be small in Asia,(10–12,22) grade 3–4 neutropenia developed in all of our patients with the UGT1A1*28/*28 genotype, and one patient died because of treatment-related sepsis. The AUC0–24 h of SN-38 was also very high in these patients. Therefore, the UGT1A1*28/*28 genotype is an important determinant of safety, even in Asian patients.

Several studies have addressed the issues of UGT1A1 polymorphisms and the starting dose of irinotecan,(11,13,22) but most of these studies had limitations, such as small numbers of patients with the rare variant alleles, a retrospective design, or the inclusion of patients with various types of cancer. Even prospective studies have included patients who received heterogeneous treatments including irinotecan.(9,10,24) We focused on high-risk patients with relatively rare UGT1A1 alleles to delineate the important pharmacogenetic determinants of irinotecan-induced neutropenia. To eliminate the potential effects of confounders such as diagnosis and concurrent therapy, we only included patients with gastric or colorectal cancer who received irinotecan monotherapy. We also evaluated the relationship between genotype and the pharmacokinetics of irinotecan and its major metabolites. We believe that these features of our study enhance the validity of our findings. The CRM used in our study offers important advantages over the conventional design with three patient cohorts.(25) We could enroll the patients promptly, and treat approximately 80% of patients at the MTD of irinotecan (150 mg/m2), as initially estimated.

A previous dose-finding study suggested that the recommended dose of 180 mg/m2 irinotecan in the FOLFIRI regimen was too low in patients with metastatic colorectal cancer who had the UGT1A1*1/*1 or UGT1A1*28/*1 genotype.(19) The incidence of grade 3–4 neutropenia was 24% in that study. In our study, the incidence of grade 3–4 neutropenia was 23% across the wild-type and heterozygous groups, supporting these earlier findings.

In the present study, only 25% of the patients in the homozygous group were able to complete two cycles of treatment at a dose of 150 mg/m2 without treatment delays and dose reduction. Previous studies(12,20) revealed that low-dose irinotecan (100–125 mg/m2) carried a low risk of neutropenia, even in patients with a UGT1A1 homozygous genotype. Although the results of a meta-analysis(21) reported an increased risk of neutropenia not only at medium or high doses of irinotecan, but also at low doses in homozygous group, the results of the present study supported the report of Hoskin et al.(20) as, similarly, no DLT occurred at doses of 100 or 125 mg/m2. This might be because retrospective studies were included and there was heterogeneity (such as cancer type, therapeutic line, regimen) in the meta-analysis,(21) whereas the present study was a prospective study using a single drug in a homogenous population of patients with colorectal cancer and gastric cancer. Moreover, the patients who required dose reduction from 150 to 125 or 100 mg/m2 were able to receive subsequent treatment safely. Therefore, UGT1A1 genetic polymorphism testing is useful, because a risk attributable to CPT-11 could be avoided by selecting another therapy, even in homozygous patients with a high risk of side-effects who have a poor performance status or a history of intensive treatment. Irinotecan at doses >150 mg/m2 has been used in regimens such as cetuximab plus irinotecan (350 mg/m2)(26) and FOLFIRI (180 mg/m2) in colorectal cancer.(27) Our results suggest that starting treatment at such high doses of irinotecan would be very risky in patients who have two alleles of UGT1A1*28 and/or UGT1A1*6. The UGT1A1 genetic polymorphism is a solid factor that affects the pharmacokinetics and a factor for judging the risk of hemotoxicity. Because CPT-11 is used as the second- or third-line rather than the first-line therapy in patients with gastric cancer and colorectal cancer, it would be better to carry out genetic testing before therapy commenced and after a full explanation to all patients. Patients who have two alleles of UGT1A1*28 and/or UGT1A1*6 can receive irinotecan at a starting dose of 150 mg/m2 and must be closely observed by carrying out observations and blood tests weekly, at least during the first cycle.

On the basis of the results obtained in our study, a nationwide close observational study is now ongoing to evaluate the safety and efficacy of irinotecan at a dose of ≤150 mg/m2 after genetic testing.

Acknowledgments

This study was funded by Yakult Honsha Co., Ltd. We thank all of the patients and investigators who participated in this study. We are also indebted to Yutaka Ariyoshi, Kazuo Tamura, Masashi Fujii, Masanori Terashima, Hironobu Minami, and Yoshinori Hasegawa for their helpful advice.

Disclosure Statement

The authors have no conflicts of interest to declare.

Abbreviations
AUC

areas under the time–concentration curves

CI

confidence interval

CRM

continual reassessment method

DLT

dose-limiting toxicity

FOLFIRI

folinic acid/fluorouracil/irinotecan

MTD

maximum tolerated dose

SN-38

7-ethyl-10-hydroxycamptothecin

SN-38G

SN-38 glucuronide

UGT

uridine diphosphate glucuronosyltransferases

Ancillary