Japanese genome-wide association study identifies a significant colorectal cancer susceptibility locus at chromosome 10p14.

Genome-wide association studies are a powerful tool for searching for disease susceptibility loci. Several studies identifying single nucleotide polymorphisms (SNP) connected intimately to the onset of colorectal cancer (CRC) have been published, but there are few reports of genome-wide association studies in Japan. To identify genetic variants that modify the risk of CRC oncogenesis, especially in the Japanese population, we performed a multi-stage genome-wide association study using a large number of samples: 1846 CRC cases and 2675 controls. We identified 4 SNP (rs7912831, rs4749812, rs7898455 and rs10905453) in chromosome region 10p14 associated with CRC; however, there are no coding or non-coding genes within this region of fairly extensive linkage disequilibrium (a 500-kb block) on 10p14. Our study revealed that the 10p14 locus is significantly correlated with susceptibility to CRC in the Japanese population, in accordance with the results of multiple studies in other races.

The rates of mortality and morbidity from colorectal cancer (CRC) have been increasing exponentially in Japan, and CRC is considered a national problem that needs to be solved urgently. Identification of factors involved in the carcinogenesis and progression of CRC would help prevent the occurrence of CRC, as well as improve the clinical outcome of treatment of the disease. For the last ten years, several genome‐wide association studies (GWAS) have identified single nucleotide polymorphisms (SNP) intimately connected to the onset of CRC. Tomlinson et al. identified rs6983267 at 8q24.21 as the SNP most strongly connected to the onset of CRC.1, 2 Zanke et al.3 investigated 100k SNP in 7480 cases of CRC with double screening among different races and discovered SNP at 8q24 as well as one at 9q24. Brodelick et al.4 report the SNP rs4938827 at 18q21 located within the gene SMAD7 from among 550k SNP in 7473 cases of CRC. Tenesa et al.5 report the SNP rs3802842 at 11q23, rs7014346 at 8q24 and rs4939827 at 18q21. Table 1 shows the data derived from previous GWAS on CRC,1, 2, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 which reveal that risky genetic polymorphisms are different among various populations. For example, in the 8q23–24 region, rs6983267 is a risk factor for CRC among Caucasians, Asians and Africans, rs7014346 and rs10505477 are risky only among Caucasians, and rs16892766 is a risk factor for those with Caucasian and African ancestry.24

Table 1

Summary of previously reported single nucleotide polymorphisms (SNP) associated with colorectal cancer

SNP	Chromosome	Gene	Odds ratio	P‐value	Population	First author	Journal	Year	Reference number
rs4939827	18q21	SMAD7	1.15	1.0 × 10(−12)	Caucasian	Broderick P	Nat Genet	2007	4
rs6983267	8q24	–	1.17	3 × 10(−11)	European	Tomlinson IP	Nat Genet	2007	1
rs4444235	14q22.2	BMP4	1.11	8.1 × 10(−10)	European	Houlston RS	Nat Genet	2008	6
rs9929218	16q22.1	CDH1	0.91	1.2 × 10(−8)	European	Houlston RS	Nat Genet	2008	6
rs10411210	19q13.1	RHPN2	0.87	4.6 × 10(−9)	European	Houlston RS	Nat Genet	2008	6
rs961253	20p12.3	–	1.12	2.0 × 10(−10)	European	Houlston RS	Nat Genet	2008	6
rs4779584	15q13	CRAC1	1.26	4.44 × 10(−14)	Ashkenazi Jews and Europeans	Jaeger E	Nat Genet	2008	7
rs3802842	11q23	–	1.1	5.8 × 10(−10)	European, Japanese, Israeli	Tenesa A	Nat Genet	2008	5
rs10795668	10p14	–	1.12	2.5 × 10(−13)	European	Tomlinson IP	Nat Genet	2008	2
rs16892766	8q23.3	EIF3H	1.26	3.3 × 10(−18)	European	Tomlinson IP	Nat Genet	2008	2
rs7136702	12q13.13	–	1.06	4.02 × 10(−8)	European	Houlston RS	Nat Genet	2010	8
rs6687758	1q41	–	1.09	2.27 × 10(−9)	European	Houlston RS	Nat Genet	2010	8
rs6691170	1q41	–	1.06	9.55 × 10(−10)	European	Houlston RS	Nat Genet	2010	8
rs4925386	20q13.33	LAMA5	0.93	1.89 × 10(−10)	European	Houlston RS	Nat Genet	2010	8
rs10936599	3q26	MYNN	0.93	3.39 × 10(−8)	European	Houlston RS	Nat Genet	2010	8
rs7758229	6q26	SLC22A3	1.28	7.92 × 10(−9)	Japanese and Korean	Cui R	Gut	2011	9
rs1957636	14q22	BMP4	1.084	3.93 × 10(−10)	Caucasian	Tomlinson IP	Plos Genet	2011	11
rs11632715	15p	GREM1	1.116	2.30 × 10(−10)	Caucasian	Tomlinson IP	Plos Genet	2011	11
rs16969681	15p	GREM1	1.181	5.33 × 10(−8)	Caucasian	Tomlinson IP	Plos Genet	2011	11
rs4813802	20p12	BMP2	1.093	4.65 × 10(−11)	Caucasian	Tomlinson IP	Plos Genet	2011	11
rs3824999	11q13.4	POLD3	1.08	3.65 × 10(−10)	European and Japanese	Dunlop MG	Nat Genet	2012	10
rs1321311	6p21	CDKN1A	1.1	1.14 × 10(−10)	European and Japanese	Dunlop MG	Nat Genet	2012	10
rs5934683	Xp22.2	SHROOM2	1.07	7.30 × 10(−10)	European and Japanese	Dunlop MG	Nat Genet	2012	10
rs4813802	20p12	BMP2	1.18	7.3 × 10(−5)	Caucasian	Peters U	Hum Genet	2012	12
rs2853668	5p33.15	TERT‐CLPTM1L	0.85	1.9 × 10(−4)	Caucasian	Peters U	Hum Genet	2012	12
rs10774214	12p13.32	CCND2	1.17	3.06 × 10(−8)	East Asian and European	Jia WH	Nat Genet	2013	13
rs2423279	20p12.3	HAO1, PLCB1	1.14	6.64 × 10(−9)	East Asian and European	Jia WH	Nat Genet	2013	13
rs647161	5q31.1	PITX1	1.17	1.22 × 10(−10)	East Asian and European	Jia WH	Nat Genet	2013	13
rs3987	4q26	NDST3	1.36	4.02 × 10(−8)	Spanish	Real LM	PLos One	2014	14
rs35509282	4q32.2	FSTL5	1.53	8.2 × 10(−9)	Ashkenazi Jews and Europeans	Schmit SL	Carcinogenesis	2014	15
rs12241008	10q25	VTI1A	1.19	1.4 × 10(−9)	European, African and Japanese	Wang H	Nat Commun	2014	16
rs1035209	10p24.2	MRP2	1.13	4.54 × 10(−11)	East Asians in our European	Whiffin N	Hum Mol Genet	2014	17
rs3217810	12p13.32	CCND2	1.19	2.16 × 10(−10)	East Asians in our European	Whiffin N	Hum Mol Genet	2014	17
rs10911251	1q25.3	LAMC1	1.09	1.75 × 10(−8)	East Asians in our European	Whiffin N	Hum Mol Genet	2014	17
rs7229639	18q21.1	SMAD7	1.22	2.93 × 10(−11)	East Asians	Zhang B	Int J Cancer	2014	18
rs704017	10q22.3	ZMIZ1‐AS1	1.1	2.07 × 10(−8)	East Asians	Zhang B	Nat Genet	2014	18
rs11196172	10q25.2	TCF7L2	1.14	1.04 × 10(−12)	East Asians	Zhang B	Nat Genet	2014	18
rs1535	11q12.2	FADS2	1.15	8.21 × 10(−20)	East Asians	Zhang B	Nat Genet	2014	18
rs174537	11q12.2	MYRF	1.16	9.22 × 10(−21)	East Asians	Zhang B	Nat Genet	2014	18
rs174550	11q12.2	FADS1	1.15	1.58 × 10(−19)	East Asians	Zhang B	Nat Genet	2014	18
rs4246215	11q12.2	FEN1	1.15	7.65 × 10(−20)	East Asians	Zhang B	Nat Genet	2014	18
rs10849432	12p13.31	CD9	1.14	5.81 × 10(−10)	East Asians	Zhang B	Nat Genet	2014	18
rs12603526	17p13.3	NXN	1.1	3.42 × 10(−8)	East Asians	Zhang B	Nat Genet	2014	18
rs1800469	19q13.2	TGFB1	1.09	1.17 × 10(−8)	East Asians	Zhang B	Nat Genet	2014	18
rs2241714	19q13.2	B9D2	1.09	1.36 × 10(−8)	East Asians	Zhang B	Nat Genet	2014	18
rs10904849	10p13	CUBN	1.14	7.01 × 10(−8)	European	Al‐Tassan NA	Sci Rep	2015	19
rs17836917	17q12	MYO1D, CCL8, CCL13	0.75	4.55 × 10(−8)	Han Chinese	Jiang K	Oncotarget	2015	20
rs12522693	5q23.3	HINT1	1.31	2.08 × 10(−8)	Han Chinese	Jiang K	Oncotarget	2015	20
rs17094983	14q23.1	–	0.87	2.5 × 10(−10)	African	Lemire M	Hum Genet	2015	21
rs16941835	16q24.1	RP11‐58A18	1.15	5.06 × 10(−8)	African	Lemire M	Hum Genet	2015	21
rs72647484	1q36.2	CDC42, WNT4	1.21	1.21 × 10(−8)	African	Lemire M	Hum Genet	2015	21
rs1119016410	10p24.1	SLC25A28, ENTPD7, COX15, CUTC, ABCC2	1.09	4.0 × 10(−8)	European and Asian	Schumacher FR	Nat Commun	2015	22
rs318450412	12q24.12	SH2B3	1.09	1.7 × 10(−8)	European and Asian	Schumacher FR	Nat Commun	2015	22
rs7320812012	12q24.22	NOS1	1.16	2.8 × 10(−8)	European and Asian	Schumacher FR	Nat Commun	2015	22
rs606682520	20q13.13	PREX1	1.09	4.4 × 10(−9)	European and Asian	Schumacher FR	Nat Commun	2015	22
rs8124813	3p14.1	LRIG1	1.09	2.0 × 10(−8)	European and Asian	Schumacher FR	Nat Commun	2015	22
rs35360328	3p22.1	CTNNB1	1.14	3.1 × 10(−9)	European and Asian	Schumacher FR	Nat Commun	2015	22
rs4711689	6p21.1	TFEB	1.11	3.9 × 10(−8)	East Asian	Zeng C	Gastroenterology	2016	23
rs2450115	8q23.3	EIF3H	1.12	1.2 × 10(−12)	East Asian	Zeng C	Gastroenterology	2016	23
rs6469656	8q23.3	EIF3H	1.11	2.0 × 10(−11)	East Asian	Zeng C	Gastroenterology	2016	23
rs4919687	10q24.3	CYP17A1	1.14	7.8 × 10(−12)	East Asian	Zeng C	Gastroenterology	2016	23
rs11064437	12p13.3	SPSB2	1.12	4.5 × 10(−11)	East Asian	Zeng C	Gastroenterology	2016	23

As for the Japanese population, several SNP‐based GWAS have been performed for CRC. Matsuo et al.25 performed a case‐control study using 481 cases and 962 controls and report an association between CRC and rs6893267 at 8q24. Furthermore, Cui et al.9 performed GWAS using 1583 Japanese CRC cases and 1898 controls and replication analyses using a total of 4809 CRC cases and 2973 controls, including 225 Korean subjects with distal colon cancer and 377 controls. They found an association between CRC and a known carcinogenic SNP at 8q24 and an association between distal CRC and rs7758229 in intron 5 of SLC22A3 at 6q26.9 Zhang et al.26 performed a case‐control study and reported that microsomal glutathione S‐transferase 1 (MGST1) gene polymorphisms had an association with CRC risk (OR = 1.682, P = 0.004) among the Han Chinese. However, the molecular biological mechanisms by which these SNP are involved in colorectal carcinogenesis remain unclear.

Using a case‐control study on 1511 CRC cases and 2098 controls, we previously reported that the risk allele of rs6983267 in 8q24 is associated with CRC, especially in diabetes mellitus patients.27 However, these SNPs, except for those in 8q24, have not been defined as the regulating polymorphisms of colorectal carcinogenesis beyond racial differences. To find the responsible host genetic factors, we designed an SNP‐based GWAS to identify SNP associated with susceptibility to morbidity from CRC in a pure Japanese population.28 We performed a multistage genome‐wide association study in Japanese individuals, with a total of 1846 cases and 2675 controls, to identify disease susceptibility SNP.

Materials and Methods

Study samples

We collected peripheral blood samples from nine collaborating institutes and hospitals for this project investigating the genetic risk factors of CRC cases. Newly diagnosed CRC cases were identified in eight hospitals (Kyushu University Beppu Hospital [Beppu, Japan], Kitazato University [Kanagawa, Japan], National Cancer Center Hospital [Tokyo, Japan], Northern Yokohama Hospital Showa University [Kanagawa, Japan], National Defense Medical College Hospital [Saitama, Japan], Tokyo Medical and Dental University Hospital [Tokyo, Japan], Mie University Hospital [Mie, Japan] and Takano Hospital [Kumamoto, Japan]) from 2000 to 2007. Controls without a prior history of CRC at the time of enrollment were also recruited from those hospitals. All controls were enrolled after having a colonoscopy to ensure that they had no disease. All participants provided documented informed content. The study protocol was reviewed and approved by each institute. We included 1846 cases and 2675 controls into the GWAS study. The average age of CRC patients was 61.9 ± 11.0 years, and that of controls was 59.8 ± 15.0 years. Age and gender details for each phase are shown in Table S1. All cases and controls were of East Asian ancestry and from Japan.

Extraction of genomic DNA and PCR of markers

Genomic DNA was extracted from samples using conventional methodologies and quantified using PicoGreen (Invitrogen, Carlsbad, CA, USA). PCR was done using GeneAmp PCR System 9700 (Applied Biosystems, Carlsbad, CA, USA). Genotyping was done using the ABI 3100 Genetic Analyzer (Applied Biosystems) and analyses and assignment of marker alleles were done with the GENOTYPER programs (Applied Biosystems). Information on SNP was obtained from the dbSNP database (http://www.ncbi.nlm.nih.gov/SNP/index.html).

Genotyping

In phase 1, the genotyping of the 577 CRC cases and 571 controls was carried out using the Affymetrix GeneChip Human Mapping 500K Array Set according to the manufacturer's protocols. The equal number of patient and control samples enabled us to analyze the genotype and phenotype independently.

In phase 2.1 (fast‐track second screening), among the whole array of SNP, we focused on the highest‐ranked 100 SNP. Among those 100 SNP, we excluded SNP with a minor allele frequency (MAF) <0.1 and selected TagSNP to avoiding overlapping and allelic imbalance, thus totaling 62 SNP that were confirmed by PCR in phase 2.1 by screening with a subset of 1181 cases and 1617 controls.

In phase 2.2 (full second screening), 480 CRC and 480 control samples were genotyped at the 1536 best SNP (allelic P < 0.013) using the Illumina GoldenGate Assay. When multiple SNP displayed strong linkage disequilibrium (LD) with each other (r2 > 0.8), the most closely associated SNP was chosen to avoid redundancy during the selection of the 1536 SNP. The samples with a genotype call rate <0.98 and SNP with a call rate <0.98, in Hardy–Weinberg disequilibrium (P < 1.0 × 10−4) in the controls, or with a MAF <0.05 were excluded from the association analysis.

For the 21 SNP that showed an allelic P < 0.01 in phase 2.2, genotyping with the TaqMan method in 789 CRC cases and 1624 controls was performed in phase 3.

Statistical analysis

Genotype data cleaning and pairwise identity‐by‐descent (IBD) analysis were carried out using the PLINK software (version 1.06). We used the Haploview software (v3.2) to establish the LD structure on chromosome 10p14.

Results

An overview of the current study design is shown in Figure 1. The genome‐wide association study was carried out using the Affymetrix GeneChip Human Mapping 500K Array Set. In phase 1, we genotyped 500 568 tagSNP in 577 individuals with CRC and 571 controls using the Bayesian robust linear model with Mahalanobis (BRLMM) algorithm (http://media.affymetrix.com/support/technical/whitepapers/brlmm_whitepaper.pdf). Samples with a genotype call rate <0.94 for either the NspI or StyI GeneChip SNP were removed from analysis. To detect duplicate samples, relatives and DNA‐contaminated samples, pairwise IBD estimation was carried out. After applying the strict quality control criteria described above, genotype data for 529 cases and 521 controls were chosen for analysis. SNP were removed from the analysis if they had a call rate of less than 0.95, showed a difference in call rate of more than 0.03 between CRC and controls, displayed Hardy–Weinberg disequilibrium (P < 1.0 × 10−6) in the control group, or had a MAF <0.10. SNP that were not selected in the updated GeneChip SNP5.0 (Affymetrix) were also excluded. After these exclusions, 280 972 SNP remained in the first stage. Figure 2a shows a Manhattan plot of these data. The genomic inflation factor based on the median χ2‐value was 1.100 in this genome‐wide association analysis (Fig. 2b), implying that there was no systematic increase of false positives owing to population stratification or any other form of bias. The strongest associations identified in phase 1 were found at the polymorphic sites rs13000465 (P = 5.16 × 10−5) on chromosome 2 and equally at rs7912831 (P = 6.57 × 10−5) on chromosome 10.

Figure 1

The complete design of the genome‐wide association study. In phase 1, 577 patients with colorectal cancer (CRC) and 571 controls were genotyped for 500 568 single nucleotide polymorphisms (SNP) with Affymetrix 500 K chip sets. Two additional rounds of screening using the Illumina GoldenGate Assay (1536 SNP for phase 2.2) and TaqMan Assay (21 SNP for phase 3) were performed to identify significant SNP.

Figure 2

(a) Manhattan plots from the phase 1 genome‐wide association results. P‐values (−log10 P, y‐axis) are plotted against their respective chromosomal positions (x‐axis). Each chromosome is depicted in alternating blue and red. (b) Log quantile‐quantile P‐values between the expected (theoretical) P‐value and the observed P‐value. The genomic inflation factor (based on median χ2) is 1.10. If we set the odds ratio of the CRC‐related genotype as 1.4 and the allelic frequency in the control as 0.2, it will be located at the 36th quantile by the P‐value distribution. If we set the odds ratio of the CRC‐related genotype as 1.4 and the allelic frequency in the control as 0.4, it will be located at the 350th. If we set the odds ratio of the CRC‐related genotype as 1.2 and the allelic frequency in the control as 0.2, it will be located in greater than the 1000th, and the genotype will be difficult to determine.

In phase 2.1, we focused on the 100 highest‐ranked SNP among the whole set of SNP. Among these 100 SNP, 62 SNP were confirmed by PCR after screening with a subset of 1181 cases and 1617 controls (Table 2). We confirmed the susceptibility of 4 SNP (rs13000465, rs7597875, rs7912831 and rs234588) to morbidity from CRC. Eventually, we identified the colorectal cancer susceptibility locus rs7912831 on chromosome 10p14 (P = 9.31E−08) (Table 2). Then, we proceeded to the phase 2.2 full screening and genotyped 480 CRC cases and 480 controls using the Illumina GoldenGate Assay for the top‐ranking 1536 SNP. A total of 21 SNP were identified in phase 2.2, including rs7912831 on chromosome 10p14.

Table 2

Fast tracked second screening (phase 2.1) of single nucleotide polymorphisms (SNP) related to colorectal cancer

Chromosome	SNP	Phase 1	Phase 2.1	First PCR	Second PCR	Total screening
1	rs325914	4.86E−05	8.08E−01
1	rs1510310	1.24E−04	5.82E−01
1	rs1442459	4.32E−05	8.54E−01
1	rs6692131	1.24E−04	2.75E−01
2	rs7586098	2.25E−04	4.17E−01
2	rs10178331	2.63E−05	7.25E−01
2	rs1036069	9.97E−06	7.52E−01
2	rs12105972	2.13E−04	5.61E−01
2	rs13000465	5.16E−05	5.09E−02	5.43E−01
2	rs12624259	2.00E−05	3.16E−01
3	rs3771021	2.84E−04	6.82E−01
3	rs7597875	1.88E−04	3.58E−02	8.13E−01
3	rs2881606	8.55E−05	4.95E−01
3	rs6794054	1.45E−04	8.50E−01
3	rs9823024	1.63E−04	9.13E−01
4	rs12491172	6.77E−05	3.80E−01
4	rs1436656	1.26E−04	2.34E−01
4	rs11725389	1.13E−04	8.91E−01
5	rs4512014	1.82E−04	2.90E−01
5	rs13112145	6.74E−05	3.20E−01
5	rs250222	2.15E−04	7.35E−01
6	rs6595624	2.65E−05	1.69E−01
6	rs2864246	1.73E−04	1.62E−01
6	rs900402	1.70E−04	5.04E−01
6	rs2523865	8.50E−05	1.67E−01
7	rs220687	1.45E−04	1.71E+00
7	rs2040432	3.83E−05	4.80E−01
8	rs10242940	1.38E−04	6.46E−01
9	rs6975879	9.73E−05	2.84E−01
10	rs3107548	8.50E−05	9.40E−01
10	rs7041802	8.59E−05	2.09E−01
10	rs11251410	7.79E−05	1.81E−01
11	rs7912831	6.57E−05	7.10E−02	1.25E−04	2.42E−05	9.31E−08
11	rs11239278	2.14E−04	5.73E−01
11	rs5030317	8.75E−05	7.18E−01
11	rs11032820	2.57E−04	4.90E−01
12	rs7124825	8.12E−05	9.79E−01
13	rs22576154	1.59E−04	6.54E−01
13	rs11068349	1.79E−04	3.16E−01
13	rs7327880	2.75E−06	8.87E−01
13	rs8002855	5.18E−05	4.41E−01
13	rs9544316	7.94E−06	8.24E−01
13	rs1329338	1.80E−04	4.27E−01
13	rs2803215	3.57E−05	7.87E−01
13	rs9577345	1.12E−04	1.51E−01
14	rs6573776	1.72E−04	8.25E−01
14	rs234588	2.14E−04	2.17E−02	2.33E−01
15	rs539901	5.63E−05	1.72E−01
16	rs11860241	3.63E−05	1.44E−01
16	rs150073	1.43E−04	7.16E−01
16	rs8047051	2.17E−04	1.79E−01
16	rs1110560	1.61E−04	6.57E−01
17	rs7214294	1.04E−04	2.65E−01
18	rs9303936	1.56E−04	2.23E−01
18	rs9957443	1.19E−04	5.47E−01
18	rs11082969	1.84E−04	8.56E−01
18	rs2879526	1.13E−04	2.52E−01
19	rs12609781	9.21E−05	9.59E−01
19	rs326444	7.44E−05	4.48E−01
20	rs736232	4.30E−06	6.78E−01
21	rs2825545	1.93E−04	7.94E−01
22	rs4822015	7.76E−05	4.01E−01

For the 21 SNP that showed an allelic P < 0.01 in phase 2, genotyping with the TaqMan method in 789 CRC cases and 1624 controls was performed in phase 3. The average SNP call rate of these 21 SNP was 0.998. We identified 4 SNP (rs7912831, rs4749812, rs7898455 and rs10905453) with an allelic P < 0.05 on chromosome 10p14. As depicted in Figure 3, these SNP are within a region of fairly extensive LD consisting of a 500‐kb block on chromosome 10p14, which maps within 55 kb to the SNP rs10795668 reported by Tomlinson et al.2 This chromosomal region has already been reported to contain polymorphic sites; however, the current study is the first to report that variants at these specific polymorphic sites are associated with susceptibility to CRC. The final findings of our GWAS in Japanese CRC cases are shown in Table 3.

Figure 3

Linkage disequilibrium (LD) structure at 10p14. The polymorphic site rs7912831 is depicted in an LD block of 500 kb where there are coding and non‐coding genes, such as non‐coding RNA and micro RNA. The SNP rs10795668, which has been previously reported by Tomlinson et al., is located close to rs7912831. Data were analyzed using Haploview software (v3.2)

Table 3

Final Findings of genome‐wide association studies in Japanese colorectal cancer cases

Chromosome	SNP	Position	Phase 1			Phase 2.2			Phase 3			Combined			Odds ratio (95% CI)
			Risk allele frequency		P (allelic)	Risk allele frequency		P (allelic)	Risk allele frequency		P (allelic)	Risk allele frequency		P (allelic)
			Case	Control		Case	Control		Case	Control		Case	Control
10p14	rs7912831	8771261	0.60	0.53	0.00045	0.62	0.55	0.0030	0.60	0.56	0.0035	0.61	0.55	0.0000000931	1.27 (1.16–1.39)
10p14	rs4749812	8777570	0.61	0.53	0.00020	0.62	0.55	0.0030	0.60	0.56	0.0046	0.61	0.55	0.0000000721	1.27 (1.17–1.39)
10p14	rs7898455	8778914	0.68	0.61	0.00045	068	0.62	0.0085	0.67	0.63	0.0027	0.68	0.62	0.000000155	1.28 (1.17–1.40)
10p14	rs10905453	8784027	0.25	0.20	0.00972	0.25	0.20	0.0082	0.25	0.21	0.0040	0.25	0.21	0.00000501	1.27 (1.15–1.41)

SNP, single nucleotide polymorphisms.

Discussion

In this study, we identified 4 SNP that are significantly associated with morbidity from CRC at the 10p14 locus. These novel SNP are close to the variants on 10p14 described by Tomlinson et al.1 These independent whole genome‐wide association studies both found loci on 10p14 to be commonly associated with CRC. We consider this 10p14 locus to be a significant region of CRC susceptibility because it was identified in multiple studies in more than one race, including European and Japanese populations.29, 30 Recently, a significant interaction between an SNP at 10p14 (rs4143094) and processed meat consumption was also reported in CRC patients.31

We performed data mining for genes that exist in this susceptibility locus in order to investigate the mechanism by which these SNP are connected to the causes of CRC. No genes, including noncoding RNA, were found in the susceptibility locus identified in the current study using the online human genome database (UCSC Genome Bioinformatics [http://genome.ucsc.edu/]) (Fig. S1). The genes nearest to our susceptibility locus at 10p14 are GATA3 and CUGBP2. However, no significant correlations between our SNP and the expression of these genes were found by the combined SNP and cDNA expression arrays. We also sought to find new transcripts related to colorectal carcinogenesis at the 10p14 locus. First, we performed a RACE assay for each region where the SNP exist using total RNA extracted from CRC cell lines (Fig. S2), but we could not find any transcripts around all 4 SNP. Second, we found a sequence homologous to mmu‐mir‐1981, which is a murine micro RNA, just beside rs10905453. However, no transcript was found with northern blotting (Fig. S3). Moreover, we also did not find new protein‐coding or RNA genes in the susceptibility region using mRNA whole‐transcriptome analysis of 25 CRC cancer tissues. We could not find any functional genes around carcinogenic SNP in 10p14 in this study. Few reports have demonstrated the mechanism by which an SNP regulates the expression of genes or noncoding RNA to promote carcinogenesis and the development of CRC, but further study is warranted.

The mechanisms of CRC carcinogenesis caused by carcinogenic SNP, including the most common SNP rs6983267 at 8q24, are not well known because these SNP do not exist in coding regions. Tuupanen et al.32 report that the risk allele of rs6983267 is associated with microsatellite‐stable cancer, and propose that the underlying germ line genetic defect in 8q24 was a target in the somatic evolution of CRC. Interestingly, they also report that the risk allele G shows a copy number increase during CRC development and that rs6983267 affected a binding site for the Wnt‐regulated transcription factor TCF4/LEF1, which leads to the enhancement of MYC transcription in vitro and in vivo.33, 34 The abundant expression of MYC contributes to carcinogenesis and progression of CRC.27, 34, 35, 36

We consider that further analysis of such loci will enable us to understand the unknown mechanisms of colorectal carcinogenesis, including discovering new genes or noncoding RNA. For example, we reported that an SNP in miR‐146a targeting EGFR and IRAK1 is associated with the prognosis of gastric cancer patients.37 As the underlying basis of the association identified at 10p14 is presently unclear, there are no clues to explain how this region is involved in morbidity from CRC. Our data reveal that 10p14 is a colorectal cancer susceptible region for more than one racial subgroup, but further studies are warranted to find the mechanistic relationship between 10p14 and colorectal carcinogenesis.

In conclusion, using a multistage GWAS in Japanese individuals, we identified a significant genome‐wide level of association for 4 SNP on chromosome 10p14 associated with the onset of CRC.

Disclosure Statement

The authors have no conflicts of interest to declare.

Table S1. Demographic data of patients and controls in this study.

Click here for additional data file.

Fig. S1. Data mining for genes which exist in oncogenic susceptibility locus using online database.

Fig. S2. Four primer sets, sandwiching 4 single nucleotide polymorphisms at 10p14 were designed for qRT‐PCR assay.

Fig. S3. Exploration of transcripts in 10p14 region with northern blotting.

Click here for additional data file.