Introduction

Lung cancer is the leading cause of cancer death in the world [1]. Epidemiological studies have shown that lung cancer is definitely linked to tobacco smoking, especially in squamous cell carcinoma and small cell carcinoma [2]. Tobacco smoke contains dozens of carcinogenic agents that are harmful to humans [3, 4]. Genetic polymorphisms in genes that encode enzymes involved in metabolizing tobacco carcinogens may affect an individual’s risk for lung cancer [5–7].

Among such genes, UDP-glucuronosyltransferase (UGT1A1) is involved in the detoxification of benzo(a)pyrene-7,8-dihydrodiol(–), a major tobacco carcinogen [8, 9]. Despite the strong association between tobacco smoking and lung cancer, the association between UGT1A1 genotype and lung cancer has not been clarified. In this study, we examined the association between UGT1A1 genotype and lung cancer prevalence.

Patients and methods

Patients

Using a prospective cohort database system (CyberOncology®; Cyber Laboratory Inc., Tokyo, Japan) and electronic medical records, we retrieved the clinical data of 5,285 patients who underwent systemic chemotherapy at Kyoto University Hospital (Kyoto, Japan) between January 2004 and June 2014. In this cohort, 1,276 lung cancer patients and 3,996 patients with other malignancies were registered. Among them, the patients who underwent UGT1A1 genotyping were eligible for this study (Fig. 1). The patient characteristics retrieved included gender, age, smoking status (current/past/never), UGT1A1 variant (*1, *6, or *28), and the primary site of malignancy. Pathology reports for the lung cancer patients were also reviewed in the electronic medical records.

Fig. 1
figure 1

Study flow

Full size image

Genotyping

Genomic DNA was extracted from whole blood. Genotyping on UGT1A1 variants was performed using the following methods.

Genotyping method 1 (January 2004–March 2011): polymerase chain reaction (PCR) was performed using the GeneAmp PCR System 9700 (Applied Biosystems, Foster City, CA, USA) with rTaq (Takara Bio, Shiga, Japan). The purified PCR products were then sequenced with a multicapillary DNA sequencer RISA384 System (Shimadzu, Kyoto, Japan). UGT1A1*6 and *28 were genotyped [10].

Genotyping method 2 (April 2011–June 2014): genotyping analyses for UGT1A1*6 and *28 were performed using Invader® UGT1A1 assay kit (Sekisui Medical Co., Ltd., Tokyo, Japan).

Statistical analysis

Logistic regression analysis using recessive, dominant and additive models without covariates was performed to identify differences in genotype frequencies between patients with lung cancer and those with other malignancies. Fisher’s exact test was used to test the association between UGT1A1 genotype and pathological classification (squamous cell carcinoma vs other pathological types, small cell carcinoma vs other pathological types) in the lung cancer patients with a smoking history. We corrected for multiple testing using the Šidák correction [11] to control the family-wise error rate at 0.05. For logistic regression analysis, we corrected for six independent tests and the corrected significant threshold was 0.0085. For Fisher’s exact test, the corrected significant threshold was 0.025 for two independent tests. Hardy–Weinberg equilibrium was also tested using the genetics package in R (https://cran.r-project.org/web/packages/genetics/index.html). Two-tailed P values <0.05 were considered significant. All statistical analyses were performed using R version 3.11 (http://www.r-project.org).

Ethics

This study was approved by the Ethics Committee of the Kyoto University Graduate School of Medicine (E2200). All patients provided written informed consent for UGT1A1 genotyping and the use of the results for research purposes.

Results

A total of 765 patients with UGT1A1 genotyping were included in this study. Among them, 194 patients had lung cancer and 671 patients had other malignancies. Patient characteristics are summarized in Table 1. The allele frequencies of UGT1A1*28 and UGT1A1*6 were in Hardy–Weinberg equilibrium (P = 0.18 and P = 0.33, respectively) in patients with other malignancies.

Table 1 Patient characteristics
Full size table

In the recessive model, UGT1A1*28*28 genotype was significantly associated with lung cancer (OR 5.3, P = 0.0083), whereas the UGT1A1*6*6 genotype was not (OR 0.9, P = 0.79). Neither the dominant model nor the additive model showed any significant association with lung cancer (Table 2).

Table 2 Recessive, dominant, and additive model results
Full size table

All of the seven lung cancer patients with UGT1A1*28*28 had a smoking history—three patients had squamous cell carcinoma, three had small cell carcinoma, and one had adenocarcinoma (classified to other types of carcinoma in this research). Among the rest of the lung cancer patients with a smoking history, 14 had squamous cell carcinoma, 74 had small cell carcinoma, and 67 had other types of carcinoma. The ratio of squamous cell carcinoma to other histological types in lung cancer was 3:1 in patients with UGT1A1*28*28 and 14:67 in patients with other UGT1A1 genotypes. Thus, squamous cell carcinoma was significantly predominant in lung cancer patients with UGT1A1*28*28 (P = 0.024). In contrast, the ratio of small cell carcinoma to other histological types did not differ according to genotype (P = 0.62).

Discussion

To our knowledge, this is the first report of a significant association between UGT1A1*28*28 genotype and lung cancer. In our study, UGT1A1*28 genotype was associated with lung cancer only in the recessive model, whereas no significant association was identified in the dominant model or in the additive model. This finding is consistent with a previous study which reported that the level of benzo(a)pyrene-7,8-dihydrodiol(–) glucuronide formation in liver microsomes was lower in people with UGT1A1*28*28 genotype compared with those with either UGT1A1*1*28 or *1*1 genotype [8]. The allele frequency of UGT1A1*28 in our study was 0.12, which is consistent with the allele frequency in Japanese individuals (0.09–0.15 [12–14]). Interestingly, the genotype frequency of UGT1A1*28*28 was higher in our lung cancer patients (3.6 vs 1.3% [14]), while this difference did not reach statistical significance possibly because of the limited sample size.

By contrast, no significant association was observed between UGT1A1*6 polymorphism and lung cancer in any of the models used in our analysis. Although UGT1A1*28 is a TATAA box polymorphism in the promoter region that results in 70% reduction in expression levels [15–17], UGT1A1*6 is a single-nucleotide polymorphism in the exon 1 region, which causes reduced bioactivities by 32% among homogenous individuals [18]. Our results may suggest that the reduced expression levels of enzyme are a more important factor for the risk of lung cancer than reduced bioactivities.

In this study, squamous cell carcinoma was significantly predominant in the lung cancer patients with UGT1A1*28*28 genotype and a smoking history. This finding is consistent with the results of a meta-analysis which showed that squamous cell carcinoma is more closely linked to tobacco carcinogens than is large cell carcinoma or adenocarcinoma [2]. These data indicated that squamous cell carcinoma may be more susceptible to reduced UGT1A1 enzyme activities toward tobacco carcinogens, such as benzo(a)pyrene-7,8-dihydrodiol(–).

Previous genome-wide association studies (GWAS) of lung cancer identified several susceptible genetic loci [19, 20]. However, UGT1A1*28 has not been identified in these previous report by GWAS, because the commonly used GWAS panels did not include the repeat variants such as UGT1A1*28. In this study, by targeting UGT1A1 polymorphisms with the use of a hospital database, we revealed the association between UGT1A1*28 polymorphism and lung cancer.

Our study has the following limitations. First, our results are obtained among from cancer patients without healthy controls, although the allele frequency of UGT1A1*28 was compatible with previous reports as discussed above [12–14]. Second, because our study was carried out at a single institution in Japan, the results should be validated in another cohort or other ethnic groups.

In summary, our study revealed the risk of lung cancer by UGT1A1*28 genotype and tobacco smoking.