1. Introduction
There were approximately 83,730 new cases of bladder cancer (64,280 in men and 19,450 in women) and approximately 17,200 deaths from bladder cancer (12,260 in men and 4940 in women) in the United States in 2021. The rates of new bladder cancer and death due to bladder cancer have been decreasing slightly in women in recent years, whereas in men, incidence rates have been decreasing, but death rates are stable [
1]. In Japan, 23,230 cases (17,555 in men and 15,675 in women) of bladder cancer were diagnosed in 2018, and the number of deaths from bladder cancer was 8911 (6014 in men and 2897 in women) in 2019 [
2].
Muscle invasive bladder cancer requires highly invasive treatments such as radical cystectomy and systemic chemotherapy. In addition, even non-muscle infiltrating bladder cancer often recurs in the bladder and requires multiple treatments. Thus, medical treatment for bladder cancer requires a great deal of time and medical expenses.
It is well known that smoking is a risk factor for developing bladder cancer [
3,
4]. Regarding alcohol drinking, the American Society of Clinical Oncology stated in 2018 that more than 5% of new cancer cases were due to alcohol consumption [
5]. We also reported that alcohol consumption is an independent risk factor for the development of bladder cancer in the Japanese population [
6].
Occupational and environmental factors are important, in addition to genetic predisposition for bladder cancer. There was a 45-year observational study in Nordic countries on the association between occupation and the development of bladder cancer. According to this study, occupations with a significantly increased incidence of urothelial cancer, with a standardized incidence ratio of 1.20 or higher, include male waiters, chimney sweeps, hairdressers, assistant nurses, seamen, plumbers, cooks and stewards, beverage workers, female tobacco workers, printers, waiters, chemical process workers, sales agents, hairdressers, mechanics, and administrators [
7].
This study examined the effects of single-nucleotide polymorphisms (SNPs) in the germline genome on the development of bladder cancer in Japan, adding occupational and industrial history as regulators.
A genome-wide association study (GWAS) comprehensively searches the entire genome for gene polymorphisms that exhibit significant frequency differences between an unrelated patient population of a specific disease and an unrelated control population.
In genome-wide studies that analyzed the genomes of bladder cancer patients, 57 SNPs that may increase susceptibility to bladder cancer were identified in the GWAS Catalog (
Supplementary Table S1). In addition, there are many GWAS papers on bladder cancer [
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20,
21,
22,
23,
24,
25]. In particular, the NAT2 slow acetylator and GSTM1 null genotype are considered to be potential genetic risk factors for the development of bladder cancer [
8]. Polymorphisms in the NAT2 gene were also investigated in Japan, with a risk ratio of 7.80-times [
26]. In addition, a relatively large GWAS for Japanese bladder cancer patients was announced in 2015, and although smoking has been examined and adjusted as an environmental factor, occupational factors have not been examined [
21]. Therefore, it is important to examine the relationship between bladder cancer in the Japanese population and SNPs by adjusting for the industrial/occupational history, in addition to sex, smoking history, and alcohol drinking history.
2. Materials and Methods
The genome was purified using 10 mL of blood mixed with EDTA collected from 352 bladder cancer patients (302 males, 50 females) and 434 control patients (395 males, 39 females) at Japan Organization of Occupational Health and Safety, Kanto Rosai Hospital and Tokyo Metropolitan Tama Medical Center. Control patients did not include those with upper tract urothelial cancer because bladder cancer and upper tract urothelial cancer are considered to be malignant tumors that are anatomically, histologically, and epidemiologically similar.
Occupational and environmental data were obtained from the Inpatient Clinico-Occupational Database of Rosai Hospital Group (ICOD-R), provided by the Japan Organization of Occupational Health and Safety. The ICOD-R includes an occupational history of current and past three jobs, information on smoking, and alcohol habits using interviews and questionnaires completed at the time of admission. Detailed occupational histories were coded with three-digit codes in the Japan Standard Occupational Classification and Japan Standard Industrial Classification corresponding to the International Standard Industrial Classification and International Standard Occupational Classification, respectively [
27]. The Japan Standard Occupational Classification is composed of 12 major groups, 74 minor groups, and 329 unit groups [
28], whereas the Japan Standard Industrial Classification is composed of 20 divisions, 99 major groups, 530 groups, and 1460 industries [
29]. Other clinical data were obtained from electronic medical records. Missing values exist due to omission or lack of description by patients.
2.1. A New Classification of Industry/Occupation
To create a new classification, we divided the occupations into four groups: professional, service, management, and blue-collar workers, and further divided the industries into three groups: white-collar industry, blue-collar industry, and service industry. These two kinds of groups were combined into a total of 12 (4 × 3) industry/occupation classes [
30,
31]. Using this classification, tentatively named the Zaitsu classification, we previously reported an association between occupation and the prognosis of bladder cancer [
32].
2.2. Clinical and Environmental Factor
From the clinical data, categorical variables were preliminarily analyzed by Fisher’s exact test between two or multiple groups, and continuous variables were preliminarily analyzed by the Mann–Whitney U test. Furthermore, logistic regression analysis was performed with the development of bladder cancer as the objective variable, whereas age, sex, Brinkman index (BI) classified into four groups (0: BI 0, 1: 1–399, 2: 400–799, 3: 800≤), alcohol consumption history (2 levels, yes or no), and industrial / occupational classifications of the longest-held job for each patient were explanatory variables.
The industrial/occupational classifications added to the logistic regression model here were: (a) industrial classification (20 divisions,
Supplementary Table S2); (b) the 35 major groups included in industrial classification divisions D (Construction), E (Manufacturing), and H (Transport and postal services); and (c) Occupational classification major groups (12 categories,
Supplementary Table S3). From the logistic regression models of a, b, and c, the explanatory variables related to industry/occupation were selected by the backward step-wise method using the Akaike information criterion.
2.3. Genotyping
Performed by Riken Genesis Co., Ltd. (Taito-Ku, Tokyo, Japan). Samples were genotyped using the Illumina Infinium Asian Screening Array-24 v1.0 BeadChip, which combines genome-wide coverage of East Asian populations, relevant clinical research content, and scalability for genomic screening. For quality control of samples, we excluded those with (i) a sample call rate < 0.99, (ii) a person with the lowest call rate from the pairs with a proportion IBD (identity-by-descent) > 0.1875, and (iii) outliers from Japanese clusters identified by principal component analysis using the genotyped samples and East Asians in the International Genome Sample Resource [
33] (The 1000 Genomes Project Consortium 2015). For quality control of genotypes, we excluded those with a (i) SNP call rate < 0.99 or (ii)
p-value for the Hardy-Weinberg Disequilibrium test < 0.001.
2.4. Imputation
We utilized SNP imputation for all samples under 1000 Genomes Project Phase 3 as a reference panel [
34]. We implemented the pre-phasing by Eagle [
35,
36] and imputation by Minimac3 [
37]. After imputation, we excluded SNPs with an imputation quality of R-square < 0.3.
2.5. GWAS
We conducted 6 GWAS patterns for bladder cancer development by logistic linear models using SNP dosage obtained by SNP imputation and Efficient and Parallelizable Association Container Toolbox (EPACTS) [
38]. In the association test, age, sex, smoking history (Brinkman Index, ordered category with 0 < 1 < 2 < 3 levels), alcohol consumption history (2 levels, yes or no), and several patterns of industrial/occupational classifications were added to logistic regression models. Tested industrial/occupational classifications were: (i) 1 variable with 20 levels for industrial classification divisions, (ii) selected industrial classification division(s) from 20 variables with 2 levels (yes or no) by the backward step-wise method in a logistic regression model without SNP dosage, (iii) 1 variable with 12 levels for occupational classification major groups, (iv) selected occupational classification major group(s) from 12 variables with 2 levels (yes or no) by the backward step-wise method in a logistic regression model without SNP dosage, (v) selected industrial classification major groups from 35 variables with 2 levels (yes or no) by the backward step-wise method in a logistic regression model without SNP dosage, and (vi) the Zaitsu classification. We also used only male samples for GWAS, taking into account sex differences in some occupations. We did not conduct GWAS using only females due to the small number of cases. We set the genome-wide significance level for our study at
p = 5 × 10
−8 and suggestive association level at
p = 10
−5 [
39].
2.6. Gene-Wise Analysis
For SNPs contained within 50 bp upstream and downstream of the gene regions defined in Ref Gene [
40], we calculated the genetic score (GS) [
41] as described below, and the association test between bladder cancer development and GS for each gene region was evaluated by the Burden test and SKAT-O test using EPACTS (version 3.2.6) (University of Michigan, Ann Arbor, MI, USA). In our study, we performed gene-wise analysis for 20,865 regions. We set the genome-wide significance level for our study at
p = 2.4 × 10
−6 (=0.05/20,865) and suggestive association level at
p = 10
−4. Adjusting factors in GWAS were also included in the gene-wise analysis. GWAS and gene-wise analysis were performed by StaGen Co., Ltd. (Taito-ku, Tokyo, Japan 111-0051).
Here, the GSi of an individual patient is equal to the weighted sum of the individual’s genotypes, xj (0, 1, 2), at SNPs in genei. Weights (βj) are calculated by EPACTS and M is the number of SNPs in genei.
2.7. Immunohistochemistry
The expression of gliomedin protein was examined by tissue immunostaining using paraffin-embedded bladder tumor tissue removed by transurethral resection of the bladder tumor. The antibody used was anti-GLDN (gliomedin) polyclonal antibody (26185-1-AP, Proteintech, Rosemont, IL, USA). Two independent pathologists evaluated histological staining by the immunoreactive score [
42] and individual scores were analyzed after averaging.
2.8. Study Approval
The Ethical Committee of the Japan Organization of Occupational Health and Safety approved the experiments (2018-2). All experiments were performed in accordance with relevant guidelines and regulations, including any relevant details. Written informed consent was received from patients prior to inclusion in the study.
4. Discussion
Kawasaki City, where Kanto Rosai Hospital is located, is a traditional heavy industry area adjacent to the Tokyo Metropolitan area. Fuchu City, where Tokyo Metropolitan Tama Medical Center is located, is a commercial and residential area on the outskirts of Tokyo. Therefore, workers engaged in the primary sector of industry and mining industry were limited in this study.
The reason why the industrial classification divisions G, L, and S, and the occupational classification major group F were particularly adopted as the adjusting factors in certain patterns of GWAS analysis is that these factors were selected by the backward step-wise method in the analysis of the patient background. In addition, the adjusting factors were selected from the industrial classification major groups included in the industrial classification divisions D, E, and H for men because a relatively large number of cases were included in these three divisions.
The occupations vulnerable to bladder cancer in previous reports were fairly specific and limited. Compared with these, this study mainly used the relatively rough classification of industry/occupation such as industrial classification divisions, major groups, and occupational classification major groups. Therefore, the purpose of this study was not to examine the relationship between SNPs and specific environmentally exposed substances, such as nicotine and aromatic amines, but rather to incorporate the contribution of broader industrial/occupational environmental factors, such as stress stimulation and work environment, into the development of bladder cancer as adjustment factors. Under these conditions, the gliomedin gene was detected in this study by GWAS and gene-wise analysis as a gene that may be associated with the development of bladder cancer in males.
GWAS is widely performed to replicate obtained results with other datasets. However, detailed recording of occupational/industrial history, such as ICOD-R, is not comprehensively enforced in Japan, making it difficult to replicate GWAS with occupational/industrial history as an adjusting factor. Therefore, in this study, in addition to GWAS to verify one SNP, the results were supported by performing gene-wise analysis to examine the association between a given pathological condition and a certain gene as a whole.
Kaneko et al. used ICOD-R occupational classification major groups to demonstrate that occupations with high physical activity reduced the risk of cancer [
27]. They also compared the categories included in the manufacturing industry division (Division E) of ICOD-R and noted that the incidence of ureter cancer in the electronics category is higher than that in the food manufacturing category [
43]. Therefore, adding the industrial/occupational classification to the adjusting factors of GWAS, even if it is relatively rough, is considered to be meaningful in examining the development of cancer.
The control group in this analysis included several malignant tumor diseases other than urothelial cancer. The inclusion of many cases of other malignancies in the control group of GWAS for bladder cancer is controversial. It is thought that pathways common to malignancies in general are less likely to be detected, but on the other hand, it may be more effective in order for pathways specific to bladder cancer to emerge.
The gliomedin gene encodes a protein containing olfactomedin-like and collagen-like domains. The gliomedin protein, which is present in both transmembrane and secretory forms, promotes the formation of the Node of Ranvier in the peripheral nervous system [
44]. Mutations in the gliomedin gene cause lethal congenital contracture syndrome [
45]. Autoantibodies to the gliomedin protein have also been identified in patients with multifocal motor neuropathy serotypes [
46]. An important paralog of the gliomedin gene is olfactomedin protein family [
47].
The expression of gliomedin mRNA and protein is found in the nuclei of many types of cancer cells, including urothelial cancer [
48]. The early deregulation of gliomedin during liver tumorigenesis was previously reported [
49]. The gliomedin paralog, olfactomedin 4, is a glycoprotein with an olfactomedin-domain, which is involved in numerous intracellular signaling pathways, including NF-κB, and is associated with innate immunity. Furthermore, olfactomedin 4 suppresses the development and progression of cancer [
50,
51], and tumorigenesis is observed in olfactomedin4 deficient mice [
52]. Thus, it is hypothesized that gliomedin is also involved in the development of bladder cancer by a mechanism similar to that of olfactomedin 4 in innate immunity and oncogenesis in a certain environment.
As the expression of the gliomedin protein in the nucleus of cancer cells is decreased in bladder cancer with strong nuclear atypia and infiltration tendency, it is speculated that gliomedin may act as a tumor suppressor factor in bladder cancer. As GWAS suggested a relationship between the gliomedin gene and the development of bladder cancer in men, further studies are required.