We have demonstrated that reduced function of CYP2D6 in the gefitinib cohort was associated with an increased risk of rash of grade 2 or more. No associations were observed in the erlotinib cohort between any adverse events and CYP2D6 phenotypes.
Almost all patients received appropriate supportive care in their treatment. In our study, gefitinib treatment showed different characteristics to erlotinib treatment. In a previous report without patient selection, treatment with gefitinib was not associated with a significant improvement in survival compared to placebo therapy
[4, 5]. However, erlotinib has been shown to prolong survival in unselected and EGFR wild-type patients with NSCLC after first-line or second-line chemotherapy
[6, 24]. For these reasons, gefitinib is administered to EGFR mutation-positive patients and patients with a higher rate of poor performance status (≥2).
In our study, gefitinib treatment showed different adverse events to erlotinib treatment. Compared to the side effects of erlotinib, patients treated with gefitinib had a significantly higher frequency of liver dysfunction. In the gefitinib group, the rate of liver dysfunction of all grades in our study was 45.3%, including 19.0% of grade 1, 10.5% of grade 2, 14.7% of grade 3, and 1.1% of grade 4. In the erlotinib group, the rate of liver dysfunction in our study was 21.3%, including 8.2% of grade 1, 8.2% of grade 2, 4.9% of grade 3, and 0% of grade 4. With respect to gefitinib therapy, Maemondo et al. reported a rate of 55% of all grades of increased levels of aminotransferases, and the rate of grade 3 or 4 was 21.5% in a Japanese cohort
. Mitsudomi et al. reported a rate of 70.1% of all grades and a rate of 16.1% of grade 3 or 4
. With respect to erlotinib therapy, an Asian phase III study showed a rate of 37% for all grades of increased levels of ALT, and a rate of 4% of grade 3 or 4
. Our results were similar to those found in previous gefitinib and erlotinib phase III clinical trials in Asian subjects.
Liver dysfunctions induced by gefitinib were reported in a few cases in which hepatotoxicity caused by gefitinib declined when gefitinib was changed to erlotinib. First, Kijima et al. suggested the possibility that CYP2D6 polymorphisms were related to gefitinib-induced hepatotoxicity. Their study described 3 cases with gefitinib-related hepatotoxicity whose genotypes were CYP2D6*1/*10, CYP2D6*10/*10, and CYP2D6*1/*5, with phenotypes of EM, IM and EM, respectively. Second, Takeda et al. reported a case and suggested that liver dysfunction was attributable to a gefitinib allergy on the basis of a positive drug lymphocyte stimulation test (DLST)
. In our study, the reduced function of CYP2D6 was not associated with an increased risk of liver dysfunction in the gefitinib cohort. Further analysis of the different metabolic profiles of CYP enzymes should be performed to clarify the metabolisms of gefitinib. The DLST of gefitinib may be considered of value in some patients with gefitinib-induced liver dysfunction.
The in vitro metabolism of gefitinib was investigated using human liver microsomes, and gefitinib metabolized mainly by expressed CYP3A4 produced a similar range of metabolites as liver microsomes
. When CYP3A4 function was low or inhibited by other drugs that inhibit CYP3A4, gefitinib metabolism that involves the formation of O-desmethyl-gefitinib and is determined by the CYP2D6 enzyme expressed in the liver was increased marginally
. Therefore, the CYP2D6 enzyme is important for the metabolism of gefitinib not only in patients with reduced CYP2D6 function, but also in patients with normal CYP2D6 function who take other drugs related to CYP3A4 inhibition. These patients treated with gefitinib may have severe skin rash due to decreased metabolism of gefitinib.
We evaluated 5 mutated alleles, CYP2D6*1, CYP2D6*2, CYP2D6*10, CYP2D6*14A and CYP2D6*14B, in 289 patients. The frequency of each allele was similar to those reported in previous Japanese studies. Two reports showed that the frequencies of the CYP2D6*1, CYP2D6*2, CYP2D6*5, CYP2D6*10, and CYP2D6*14 alleles were 40.1%, 12.9%, 6.2%, 38.6%, and 2.2%, respectively, in 162 Japanese
, and 43%, 12.3%, 4.5%, 38.1%, and 0.7%, respectively, in 412 Japanese
. A further report showed that the frequencies of the CYP2D6*1, CYP2D6*2, CYP2D6*5, and CYP2D6*10 alleles were 42.3%, 9.2%, 6.1%, and 40.8%, respectively, in 98 healthy Japanese
. Taken together, the average frequencies of the functional and reduced function alleles in these studies were 54.2%, and 44.7%, respectively. Asians have a high frequency of the reduced function alleles that ranges from 43% to 47%
[21–23, 28]. However, in Caucasians, the functional and reduced function alleles represented a median frequency of 71% and 26%, respectively, in all cohort studies
 and 68.8% and 25.36%, respectively, in German subjects
. Thus, Asians have a higher frequency of reduced function alleles than do Caucasians. In addition, population-related pharmacogenomics revealed a significant difference between Japanese and US patients in genomic distribution and genotype-related associations with patient outcomes for CYP3A4*1B and ERCC2. These facts may explain the frequency differences of adverse events, particularly rash, that are based on race.
CYP2D6 metabolizes many clinically important drugs, including antidepressants, neuroleptics, beta-blockers, anti-arrhythmics, and anti-cancer agents. In breast cancer patients who were treated with tamoxifen, the CYP2D6 phenotype was associated with survival
 and the concentration of the active tamoxifen metabolite, endoxifen
. Recently, in 2 single-agent studies with gefitinib in bronchioloalveolar and head and neck carcinomas, an association between the occurrence of skin toxicity and survival was found
[33, 34]. In our study, the subjects with reduced CYP2D6 function were associated with an increased risk of rash in the gefitinib cohort. Reduced CYP2D6 function may relate to longer overall survival as well as poor metabolism of gefitinib. A prospective large clinical trial is warranted to clarify these relationships.
Our findings had some limitations. The number of patients was too small to have sufficient power to detect significant differences in other adverse events between CYP2D6 phenotypes. This study was a retrospective analysis. The identification of adverse effects was prompted by monthly visits with hematological tests or a medical interview for onset of symptoms, although the hematological toxicities were in some cases diminished at the time of next visit because of self-judgment for discontinuation. The adverse events were generally controlled, except for interstitial lung disease. The relation between CYP2D6 and carcinogenic risk was not evaluated. The blood concentrations of gefitinib and erlotinib and the metabolites of gefitinib and erlotinib, were not measured. However, in clinical settings, it may be difficult to perform blood sampling at sufficient frequency to calculate area under the curve (AUC). We could not separate the UM cohort from the EM cohort. UM consists of CYP2D6*1 or CYP2D6*2, and this group was included with EM in this study. Other cytochrome P450 enzymes were not measured. The relationship between these concentrations and these enzyme phenotypes remain to be elucidated. Our study may provide useful information regarding drug selection for EGFR-TKIs. When gefitinib is administered in combination with drugs which inhibit CYP2D6 function, the frequency of severe rash by gefitinib may be increased. When the patients with unknown CYP2D6 phenotype, have a severe rash by gefitinib, the change of gefitinib to erlotinib may be sometimes useful, because the patient may have reduced CYP2D6 function and CYP2D6 is not affected the erlotinib metabolism. CYP2D6 genotyping and an understanding of CYP2D6 function may be helpful in predicting rash during gefitinib therapy.