Genetic variants in Fas signaling pathway genes and risk of gastric cancer

Abstract

Populations in north central China are at high risk for gastric cancers (GC), and altered FAS-mediated cell signaling and/or apoptosis may contribute to this risk. We examined the association of 554 single nucleotide polymorphisms (SNPs) in 53 Fas signaling-related genes using a pathway-based approach in 1758 GC cases (1126 gastric cardia adenocarcinomas (GCA) and 632 gastric noncardia adenocarcinomas (GNCA)), and 2111 controls from a genome-wide association study (GWAS) of GC in ethnic Chinese. SNP associations with risk of overall GC, GCA and GNCA were evaluated using unconditional logistic regressions controlling for age, sex and study. Gene- and pathway-based associations were tested using the adaptive rank-truncated product (ARTP) method. Statistical significance was evaluated empirically by permutation. Significant pathway-based associations were observed for Fas signaling with risk of overall GC (P = 5.5E-04) and GCA (P = 6.3E-03), but not GNCA (P = 8.1E-02). Among examined genes in the Fas signaling pathway, MAP2K4, FAF1, MAPK8, CASP10, CASP8, CFLAR, MAP2K1, CAP8AP2, PAK2 and IKBKB were associated with risk of GC (nominal P < 0.05), and FAF1 and MAPK8 were significantly associated with risk of both GCA and GNCA (nominal P < 0.05). Our examination of genetic variation in the Fas signaling pathway is consistent with an association of altered Fas signaling and/or apoptosis with risk of GC. As one of the first attempts to investigate a pathway-level association, our results suggest that these genes and the Fas signaling pathway warrant further evaluation in relation to GC risk in other populations.

INTRODUCTION

Gastric carcinoma (GC) is the fourth most common malignancy worldwide with an estimated incidence of 934,000 new cases per year. ^{1, 2} Furthermore, this incidence is geographically varied with more than 42% of GC patients occurring in China alone. ³ Globally, approximately 738,000 patients with GC die annually making GC the second most common cause of cancer-related deaths. ^{2, 4} This cancer also continues to have very poor survival, primarily because most patients present with advanced disease and treatment options are limited. ^{5, 6}

Populations from the Shanxi Province and Linxian in north central China are at very high risk for GC including gastric cardia adenocarcinoma (GCA) that arises in the top 3cm of the stomach, and gastric noncardia adenocarcinoma (GNCA), that arises more distally in the stomach. Previous studies have reported several risk factors associated with higher risk of GC in these populations including age, male gender, Helicobacter pylori (H. pylori) infection, ⁷ consumption of salted and nitrated foods, low levels of antioxidants, low consumption of fresh fruit, vegetables and eggs, ^{4, 8-10} tooth loss,¹¹ and thermal damage due to consumption of scalding hot foods. ⁴ In contrast, smoking and alcohol are not major risk factors. ^{4, 10}

In addition to environmental risk factors, data on family history of GC and genome-wide association studies ^12-14 in these high risk populations suggest the importance of genetic susceptibility. To date, five susceptibility loci at 1q22, 3q13, 5p13, 10q23 and 20p13 have reached genome-wide significance in scans conducted in Han Chinese; specifically three loci have been associated with risk of GCA and two with GNCA. ^13-15 Pathway-based analysis of genome-wide association study (GWAS) data is a complementary approach to identify pathways or groups of genes enriched with cancer associated SNPs whose individual effect sizes may be too small to be detected by standard methods.

The ability to avoid apoptosis and ensure continued proliferation and survival of premalignant and early tumor cells is likely to be an early and important event facilitating the development of cancer. Fas is a death domain-containing member of the TNFR (Tumor Necrosis Factor Receptor) superfamily and it has a central role in the physiological regulation of apoptosis. Although activated Fas (FasL-Fas system) has been appreciated mainly with respect to its death-inducing function, which is mediated via proteolytic enzymes called ‘caspases’ (CASP). ¹⁶ Fas signaling may also transduce proliferative and activating signals, through nuclear factor-kappaB (NF-kB) activation and other mechanisms.¹⁷ In mice, during early infection with H.pylori, Fas-mediated apoptosis depletes parietal and chief cell populations, leading to architectural distortion. Thus, the deregulation of FAS signaling may be an early and necessary trait for GC development and also important for H.pylori infection. ^{17, 18}

Genetic variation may alter the expression or activity of proteins in the FAS signaling pathway, potentially altering cell proliferation, apoptosis, and survival, and thus susceptibility to GC. Therefore, we evaluated 53 candidate genes associated with FAS signaling including genes downstream of Fas, initiator caspases and signal transduction effectors using ad hoc analysis of the first phase of a genome-wide association study (GWAS) of gastric cancer conducted in a high risk Chinese population. We present data here suggesting that overall Fas signaling and specific genes contained therein may be important for GC development and type of GC in high risk Chinese individuals.

METHODS & ANALYSES

Study Population

This study reports a further statistical analysis of the first phase of a genome-wide association study of GC conducted in ethnic Chinese, full details of which have been described elsewhere. ¹³ Briefly, participants for were drawn from two studies, the Shanxi Upper Gastrointestinal Cancer Genetics Project (Shanxi) and the Linxian Nutrition Intervention Trial (NIT), a prospective cohort. The Shanxi study controls were individually matched on age and sex for the case-control portion, whereas the NIT controls were selected as a case-cohort and frequency matched on age and sex. For the Shanxi and NIT studies, tumor anatomic location (cardia and noncardia) was known for all cases and >85% of cases had pathological confirmation. All GCAs were located in the proximal 3 cm of the stomach. Risk factor information for Shanxi and NIT were obtained by interview. The NCI Special Studies Institutional Review Board approved the overall GWAS.

Gene and SNP Selection for Fas Signaling Pathway

An inherent limitation of pre-processed pathway databases is the subjective interpretation of the curator. Therefore, to obtain as comprehensive a pathway as possible at the time of this study, genes associated with Fas signaling (Fas receptor and ligand, effector caspases, and downstream effectors, collectively referred to here as Fas signaling pathway genes) were identified apriori from the literature ^16-28 and cross-referenced with the Biocarta fas signaling pathway (cd95) database (BioCarta_pid_faspathway and http://cgap.nci.nih.gov/Pathways/BioCarta/h_fasPathway) to confirm pathway information. Using this approach we identified 53 genes containing 668 unique SNPs from the GWAS. The 53 genes examined in this study are listed in Table 2.

Table 2.

Pathway-, gene- and most significant SNP-based P-values for Fas-signaling pathway genes and risk of GC in China

Pathway p*	Gene Abbtn	Gene name	Cytogenetic Locus	Genepg Pg	No. of SNPs	Most Significant SNP	Associated SNP Pn
0.00055	MAP2K4	Mitogen-activated protein kinase kinase 4	17p11.2	0.0038	16	rs9788973	0.00035
	FAF1	FAS (TNFRSF6) associated factor 1	10q24.1	0.0039	28	rs1846522	0.00039
	MAPK8	Mitogen-activated protein kinase 8	10q11.22	0.0041	5	rs10508902	0.00153
	CASP10	Caspase 10, apoptosis-related cysteine peptidase	2q33-q34	0.0110	5	rs10200857	0.00382
	CASP8	Caspase 8, apoptosis-related cysteine peptidase	2q33-q34	0.0130	9	rs3769825	0.00247
	CFLAR	Capase 8 and FADD-like apoptosis regulator	2q33-q34	0.0149	5	rs10200857	0.00382
	MAP2K1	Mitogen-activated protein kinase kinase 1	15q22.1- q22.33	0.0185	9	rs12050732	0.00430
	CASP8AP2	Caspase 8 associated protein 2	6q15	0.0200	8	rs11967579	0.00409
	PAK2	p21 protein (Cdc42/Rac)-activated kinase 2	3q29	0.0476	11	rs6583176	0.00877
	IKBKB	Inhibitor of kappa light polypeptide gene enhancer in B-cells, kinase beta	8p11.2	0.0480	6	rs5029748	0.01824
	PARP1	Poly (ADP-ribose) polymerase 1	1q41-q42	0.0650	13	rs1805410	0.00863
	UBE2I	Ubiquitin-conjugating enzyme E2I	16p13.3	0.0653	4	rs8063	0.02773
	PAK1	p21 protein (Cdc42/Rac)-activated kinase 1	11q13-q14	0.0908	11	rs2725830	0.01745
	NFKB2	Nuclear factor of kappa light polypeptide gene enhancer in B-cells 2	10q24	0.1021	3	rs1056890	0.03716
	RB1	Retinoblastoma 1	13q14.2	0.1129	8	rs990814	0.03271
	PRKDC	Protein kinase, DNA-activated, catalytic polypeptide	8q11	0.1163	11	rs2213178	0.01744
	RAF1	V-raf-1 murine leukemia viral oncogene homolog 1	3p25	0.1315	18	rs904453	0.02647
	DFFB	DNA fragmentation factor, 40kDa, beta polypeptide	1p36.3	0.1334	12	rs10797348	0.01571
	CASP6	Caspase 6, apoptosis-related cysteine peptidase	4q25	0.1466	7	rs3181187	0.03006
	CASP2	Caspase 2, apoptosis-related cysteine peptidase	7q34-q35	0.1663	3	rs10500136	0.07204
	TRAF1	TNF receptor-associated factor 1	9q33-q34	0.1988	5	rsl0985097	0.07332
	TRAF2	TNF receptor-associated factor 2	9q34	0.2183	7	rs7019752	0.09501
	ARHGDIB	Rho GDP dissociation inhibitor (GDI) beta	12pl2.3	0.2253	19	rsl0505784	0.02073
	MAP3K5	Mitogen-activated protein kinase kinase kinase 5	6q22.33	0.2931	33	rs9402838	0.01781
	CASP7	Caspase 7, apoptosis-related cysteine peptidase	10q7	0.3112	19	rslll96449	0.03379
	BID	BH3 interacting domain death agonist	22qll.l	0.3137	17	rs382013	0.02841
	MAP3K1	Mitogen-activated protein kinase kinase kinase 1	5qll.2	0.3212	17	rs832585	0.05083
	MAP3K14	Mitogen-activated protein kinase kinase kinase 14	17q21	0.3411	10	rs7222751	0.06109
	APAF1	Apoptotic peptidase activating factor 1	12q23.1	0.3866	16	rs2288714	0.07195
	DIABLO	Diablo, lAP-binding mitochondrial protein	12q24.31	0.4121	2	rsl2870	0.23147
	LMNB2	Lamin B2	19pl3.3	0.4365	6	rs3729535	0.14381
	PTPN13	Protein tyrosine phosphatase, non-receptor type 13	4q21.3	0.4500	22	rs989902	0.08045
	CASP9	Caspase 9, apoptosis-related cysteine peptidase	lp36.21	0.4541	9	rs2042370	0.14994
	CASP3	Caspase 3, apoptosis-related cysteine peptidase	4q34	0.4695	2	rs2720376	0.27083
	BIRC3	Baculoviral IAP repeat containing 3	llq22	0.5177	2	rs2846848	0.34858
	BIRC5	Baculoviral IAP repeat containing 5	17q25	0.5409	11	rsl042541	0.13041
	Fas	Fas (TNFRSF6)-associated via death domain	10q24.1	0.5511	22	rsl2765241	0.10180
	BIRC2	Baculoviral IAP repeat containing 2	llq22	0.5561	2	rsl0895290	0.34386
	CRADD	CASP2 and RIPK1 domain containing adaptor with death	12q21.33-	0.5777	54	rs3858606	0.03034
		domain	q23.1
	LMNA	Lamin A/C	lq22	0.6312	4	rs915179	0.23595
	FADD	Fas (TNFRSF6)-associated via death domain	llql3.3	0.6406	4	rs481845	0.27200
	CYCS	Cytochrome C	7pl5.3	0.6612	4	rs2285738	0.28806
	RIPK2	Receptor-interacting serine-threonine kinase 2	8q21	0.6922	10	rs39765	0.19500
	SUM01	SMT3 suppressor of mif two 3 homolog 1 (S. cerevisiae)	2q33	0.7013	2	rs7599810	0.52176
	MAPK3	Mitogen-activated protein kinase 3	16pll.2	0.7641	3	rs8061772	0.43900
	DFFA	DNA fragmentation factor, 45kDa, alpha polypeptide	Ip36.3-p36.2	0.7834	5	rs2781233	0.34981
	LMNB1	Lamin B1	5q23.2	0.7837	16	rs3828699	0.24698
	CHUK	Conserved helix-loop-helix ubiquitous kinase	10q24-q25	0.8072	5	rs7073610	0.36423
	NFKB1	nuclear factor of kappa light polypeptide gene enhancer in B-cells 1	4q24	0.8118	15	rs3774937	0.22925
	JUN	Jun proto-oncogene	1p32-p31	0.8298	4	rs2760494	0.38830
	FasLG	Fas (TNFRSF6) ligand	1q23	0.8412	7	rs2859228	0.30806
	SMPD2	Sphingomyelin phosphodiesterase 2, neutral membrane	6q21	0.8893	4	rs1476387	0.49245
	DAXX	Death-domain associated protein	6p21.3	0.9446	4	rs3130267	0.72723

The pathway P-value (P*) for all 53 genes is indicated. Gene-based P-values (P^g) are shown in order of lowest to highest P-value. The most significant SNP (nonminal P-value (Pⁿ)) in each gene is indicated. Genes with P^g <0.05, are highlighted in grey. Abbreviations (Abb^tn): GC, gastric carcinoma; SNP, single nucleotide polymorphism.

Genotyping, Quality Control, and Exclusions

DNAs were genotyped as part of the GWAS at the Core Genotyping Facility of the National Cancer Institute’s Division of Cancer Epidemiology and Genetics as previously described ⁽¹³⁾. Data is available upon request from the NIH Data Access Committee (http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000361.v1.p1). An overall subject completion rate of 85% was applied to cases and controls in the combined population for all assays analyzed. We excluded SNPs with <95% completion and <95% concordance, and a minor allele frequency (MAF) <1%. After exclusion criteria were applied, 550 unique SNPs in 53 FAS signaling pathway genes remained for analysis in GC (Supplementary Table 1); 548 SNPs for GCA, and 546 for GNCA (Supplementary Tables 3-4). Linkage disequilibrium (LD) in the combined data was further computed between any two SNPs in the same gene among the combined controls using Haploview (http://www.broad.mit.edu/mpg/haploview/).

Statistical Analyses

To investigate variation in Fas signaling pathway genes and risk of GC in the GWAS data, we carried out individual SNP-, gene-, and pathway-based analyses for GCA and GNCA subtypes as well as GC overall. SNP-based analyses of each individual study as well as the combined population were tested under the additive model, and odds ratios and 95% confidence intervals were calculated using unconditional logistic regression with adjustment for age (10 year categories), sex and study in primary models. For some SNPs we used a dominant model because of the low frequency of the homozygous genotype in our population. In secondary models we also adjusted for alcohol, smoking, H.pylori and family history of UGI cancer

All P-values for SNPs are nominal except where otherwise specified. SNP-based analyses were performed using STATA version 9.0 and program language R (http://www.r-project.org/). After excluding SNPs with pairwise LD r²≥0.80 in controls, a Bonferroni-corrected threshold of P < 1.44E-04 was calculated using 345 independent SNP signals.

We conducted a gene-based analysis to evaluate the association between a candidate gene/region and cancer risk. The test statistic used was the minP statistic that was the minimum P-value among all P-values from the single SNP analysis conducted within the candidate gene. The P-value for the gene-based analysis (called gene P-value) can be evaluated through a bootstrap procedure. ²⁹ Lastly, we conducted pathway analysis to evaluate the association between the candidate genes included in the Fas signaling pathway and cancer risk. The pathway analysis was based on the ARTP method and was implemented in the R package ARTP (http://dceg.cancer.gov/bb/tools/artp). The ARTP method aims at maximizing the association signal by combining gene-level P-values from a set of selected genes within the pathway into the test statistic and uses a bootstrap procedure to estimate its P-value, and has been shown to account properly for the type I error.²⁹ The bootstrap procedure is used for the purpose of generating datasets under the null hypothesis while keeping the correlation among SNPs the same as that in the observed dataset. The P-value for both the gene-based and pathway analyses was estimated by 20,000 parametric bootstrap steps. We also considered a more stringent Bonferroni-corrected significance threshold for gene-based analysis to account for testing 53 genes (P=9.43×10⁻⁴, 0.05/53 genes).

RESULTS

Population Characteristics

In the present study we analyzed genotype data from 1,758 GC cases and 2,111 controls. Detailed characteristics and risk factors for GC in both NIT and Shanxi samples have been previously reported. ^{4, 11} A summary of demographic, risk factor, and anatomical site information for each individual study and the combined study population is shown in Table 1. In the combined population, cases were more likely to be male, drink alcohol, smoke, and have a family history of UGI cancer compared to controls. The mean age for cases of GCA, GNCA, and GC overall was higher in Shanxi compared to NIT, the proportion of male GC cases was also greater in Shanxi compared to NIT. A higher percentage of participants from the Shanxi study were ever drinkers and smokers, while participants from the NIT study had a stronger family history of UGI cancer.

Table 1.

Population characteristics in combined and individual Chinese population GWAS studies

	Combined				Shanxi				NIT
Characteristic	GNCA	GCA	Total GC	controls	GNCA	GCA	Total GC	controls	GCNA	GCA	Total GC	controls
Total, n	632	1126	1758	2111	531	864	1395	1660	101	262	363	451
Male, n (%)	458 (72.5%)	885 (78.5%)	1342 (76.3%)	1434 (67.9%)	400 (75.3%)	731 (84.6%)	1131 (81.1%)	1226 (73.9%)	58 (57.4%)	153 (58.4%)	211 (58.1%)	208 (46.1%)
Age (SD)	54.6 (10)	57.2 (9.0)	56.3 (9.5)	55.98 (9.5)	55.4 (10.4)	59.3 (8.4)	57.8 (9.4)	57.8 (9.2)	50.6 (6.6)	50.3 (7.1)	50.4 (7.0)	49.5 (7.4)
Alcohol, Yes, n (%)	122 (19.3%)	210 (18.7%)	332 (18.9%)	298 (14.11%)	120 (22.6%)	203 (23.5%)	323 (23.2%)	287 (17.3%)	2 (1.9%)	8 (3.1%)	10 (2.8%)	13 (2.4%)
Smoking, Yes, n (%)	397 (62.8%)	735 (45.3%)	1132 (64.4%)	1218 (57.7%)	357 (67.2%)	628 (72.7%)	990 (70.9%)	1076 (64.8%)	36 (35.6%)	109 (41.6%)	146 (40.3%)	142 (31.5%)
Family history UGI, Yes, n (%)	130 (20.6%)	296 (26.3%)	426 (24.2%)	478 (22.6%)	96 (18.1%)	204 (23.6%)	298 (21.4%)	338 (20.4%)	36 (35.6%)	93 (35.5%)	129 (35.5%)	140 (31.0%)

Abbreviations: GC, gastric cancer; GCA, gastric cardia adenocarcinoma; GCNA, gastric noncardia adenocarcinoma; UGI, upper gastrointestinal; SD, standard deviation; Shanxi, Shanxi Upper Gastrointestinal Cancer Genetics Project; NIT, Linxian General Population Nutrition Intervention Trial.

Fas Signaling Pathway and GC Risk

Pathway-based analysis for all 53 genes involved in Fas signaling was significantly associated with risk of GC (P = 5.5E-04) (Table 2).

Gene-based analyses identified ten genes associated with overall risk of GC (ARTP P<0.05) (Table 2) including MAP2K4 (P = 0.0038), FAF1 (P = 0.0039) MAPK8 (P = 0.0041), CASP10 (P=0.011), CASP8 (P= 0.012), CFLAR (P= 0.015), MAP2K1 (P= 0.0185), CASP8AP2 (P= 0.02), PAK2 (P= 0.0476) and IKKB (P= 0.048). P-values for the remaining 43 FAS signaling pathway genes and their most significant SNPs are shown in Table 2 and Supplementary Table 2. However, these genes did not remain significant after Bonferroni correction for multiple comparisons.

Seventy SNPs in introns and/or non-coding gene regions across 24 Fas signaling pathway genes (including: ARHGDIB, BID, CASP6, CASP7, CASP8, CASP8AP2, CASP10, CFLAR, CRADD, DFFB, FAF1, IKBKB, MAP2K1, MAP2K4, MAP3K5, MAPK8, NFKB2, PAK1, PAK2, PARP1, PRKDC, RAF1, RB1, and UBE2I) were significantly associated (P<0.05) with risk of GC in the combined population (Supplementary Table 1). The effect size and direction of SNPs were similar in both individual studies (Supplementary Table 1). After accounting for LD (r² ≥0.80), the 70 significant SNPs were shown to represent 34 independent or separate signals. We identified two SNPs in MAP2K4 and four SNPs in FAF1 that were significant at the P < 0.001 level. MAP2K4 rs9788973 (T allele) (per allele OR: 1.18, 95% CI: 1.08-1.29, P=0.0003) was shown to be in strong LD (r²≥ 0.95) with rs7216812 (C allele), which was also associated (P = 0.0005) with increased risk of GC cancer. FAF1 rs1846522 (A allele), rs7543272 (C allele), rs12089041 (T allele), and rs3789587 (T allele) were significantly associated with reduced risk of GC (Supplementary Table 1). Strong LD r² =0.96 was observed between both FAF1 rs1846522 and rs12089041, and rs1846522 and rs3789587, respectively. However, no individual SNP remained significant after Bonferroni correction for multiple comparisons.

Further adjustment for smoking, alcohol, and family history of UGI cancer did not alter these results (data not shown). H.pylori serology data were available only for NIT study participants, however, H.pylori seropositivity was essentially universal, which precluded a meaningful evaluation of the results.

Fas Signaling Pathway and Risk of GCA and GNCA

Genetic variation in the FAS signaling pathway was significantly associated with risk of GCA (P = 6.3E-03), but not GNCA (P = 8.0E-02) in our high risk population (Table 3). Gene-based analyses identified some shared susceptibility loci for both GCA and GNCA. FAF1 and MAPK8 were significantly associated with risk of both GCA (P = 0.0265 and 0.0412, respectively) and GNCA (P = 0.0436 and 0.0077, respectively) (Table 3). A number of potential cancer-specific loci were also identified between GCA and GNCA. CASP8, CASP10, CFLAR, and MAP2K1 were significantly associated with risk of GCA only (P < 0.02), while MAP2K4 and IKBKB were only significantly associated with GNCA (P < 0.05) (Table 3). However no SNP remained significant after correction for multiple comparisons. The most significant SNP in each of the 53 genes in the FAS signaling pathway for GCA and GNCA is shown in Supplementary Tables 3 and 4, respectively.

Table 3.

Pathway- and gene-based P-values for Fas-signaling pathway genes and risk of GC, GCA and GNCA in China

GC				GCA			GNCA
Pathway p*	Gene Abbtn	No. of SNPs	Gene Pg	Pathway P*	No. of SNPs	Gene Pg	Pathway P*	No. of SNPs	Gene Pg
0.00055	MAP2K4	16	0.0038	0.00634	16	0.0529	0.08054	16	0.0127
	FAF1	28	0.0039		28	0.0265		28	0.0412
	MAPK8	5	0.0041		5	0.0436		5	0.0077
	CASP10	5	0.0110		5	0.0151		5	0.1817
	CASP8	9	0.0130		9	0.0043		9	0.5739
	CFLAR	5	0.0149		5	0.0200		5	0.3181
	MAP2K1	9	0.0185		9	0.0203		9	0.4356
	CASP8AP2	8	0.0200		8	0.0779		8	0.1397
	PAK2	11	0.0476		11	0.1252		11	0.1777
	IKBKB	6	0.0480		6	0.2121		6	0.0478
	PARP1	13	0.0650		13	0.0471		13	0.2756
	UBE2I	4	0.0653		4	0.1808		4	0.0698
	PAK1	11	0.0908		11	0.1295		11	0.5396
	NFKB2	3	0.1021		3	0.1113		3	0.5882
	RB1	8	0.1129		8	0.2421		8	0.0871
	PRKDC	11	0.1163		11	0.0924		11	0.6748
	RAF1	18	0.1315		18	0.1114		18	0.5446
	DFFB	12	0.1334		12	0.0370		12	0.7200
	CASP6	7	0.1466		7	0.2131		7	0.0377
	CASP2	3	0.1663		3	0.2600		2	0.2342
	TRAF1	5	0.1988		5	0.0953		5	0.1652
	TRAF2	7	0.2183		7	0.1596		7	0.6689
	ARHGDIB	19	0.2253		19	0.1496		19	0.3283
	MAP3K5	33	0.2931		33	0.2315		33	0.5812
	CASP7	19	0.3112		19	0.5308		19	0.0542
	BID	17	0.3137		17	0.2084		17	0.7540
	MAP3K1	17	0.3212		17	0.3184		17	0.7026
	MAP3K14	10	0.3411		10	0.4773		10	0.1382
	APAF1	16	0.3866		16	0.3813		16	0.5208
	DIABLO	2	0.4121		2	0.5218		2	0.6363
	LMNB2	6	0.4365		6	0.7393		6	0.2731
	PTPN13	22	0.4500		20	0.3580		22	0.6403
	CASP9	9	0.4541		9	0.0542		9	0.8562
	CASP3	2	0.4695		2	0.3405		2	0.6957
	BIRC3	2	0.5177		2	0.5874		2	0.7775
	BIRC5	11	0.5409		11	0.7644		11	0.4939
	Fas	22	0.5511		22	0.2186		19	0.8238
	BIRC2	2	0.5561		2	0.5361		2	0.8044
	CRADD	54	0.5777		54	0.1875		54	0.7499
	LMNA	4	0.6312		4	0.3230		4	0.0897
	FADD	4	0.6406		4	0.7635		4	0.6411
	CYCS	4	0.6612		4	0.3477		4	0.6315
	RIPK2	10	0.6922		10	0.8194		10	0.0497
	SUMO1	2	0.7013		2	0.8105		2	0.6782
	MAPK3	3	0.7641		3	0.8107		3	0.8729
	DFFA	5	0.7834		5	0.6841		5	0.1002
	LMNB1	16	0.7837		16	0.8502		16	0.0926
	CHUK	5	0.8072		5	0.9289		5	0.7569
	NFKB1	15	0.8118		15	0.7686		15	0.4729
	JUN	4	0.8298		4	0.7242		4	0.9575
	FasLG	7	0.8412		7	0.9368		7	0.5939
	SMPD2	4	0.8893		4	0.6006		4	0.9842
	DAXX	4	0.9446		4	0.8978		4	0.7273

Gene-based P-values (P^g) are shown in order of lowest to highest P-value for GC in the combined population. Pathway P-value (P*) for all 53 genes in overall GC, GCA and GNCA are indicated. Genes with P^g<0.05 for GC are bolded. Color bars indicate genes commonly or differentially associated with risk of GCA and/or GNCA. Abbreviations (Abb^tn): GC, gastric cancer; GCA, gastric cardia adenocarcinoma; GNCA, gastric noncardia adenocarcinoma; SNP, single nucleotide polymorphism.

DISCUSSION

We evaluated the impact of genetic variation in the overall Fas signaling pathway with risk of GC using an ad hoc analysis of the first phase of a genome-wide association study (GWAS) of gastric cancer performed in a high risk Chinese population. The genes examined in this pathway encode proteins involved in FAS receptor-ligand binding, initiator and effector caspases, signaling, and downstream regulatory and structural proteins.

When all 53 candidate Fas signaling genes were considered, we observed a significant pathway-based association with overall GC risk (P= 5.5E-04) and GCA risk (P = 6.3E-03), but not GNCA risk (P = 8.0E-02). Furthermore, we found evidence that genetic variation in ten individual genes significantly contributed to overall GC risk in this population. In particular, FAF1 and MAPK8 were significantly associated with both GCA and GNCA risk; CASP10, CASP8, CFLAR and MAP2K1 were significantly associated with risk of GCA; and MAP2K4 and IKBKB were significantly associated with GNCA. Polymorphisms in these genes have been previously examined for risk association in a number of cancers in both Chinese and Caucasian populations (summary presented in Supplementary Table 5). However, with the exception of IKBKB rs5029748,³⁰ which was associated with a reduced risk of GC (per allele OR: 0.90; 95%CI: 0.81-0.98) and GNCA (per allele OR: 0.86; 95%CI: 0.75-0.97) in our study; we failed to replicate any of these previously-reported observations.

The lack of a pathway-based association for the Fas signaling genes with GNCA may reflect the smaller number of GNCA cases (n = 632) genotyped in this study population. Alternatively, this result may reflect differences in Fas signaling (apoptosis vs. proliferation) in the development of the GC subtypes in our high-risk Chinese population. In support of this proposal, Boroumand-Noughabi and colleagues ³¹ found a significantly higher serum level of soluble FasL in Iranian patients with GNCA versus those with GCA (P = 0.005) suggesting difference in the efficacy of apoptosis in different gastric subtype tumors and/or patient immune reponse to the subtypes. Also, other data suggests that GCA is distinguished from GNCA by differences in risk factors, ³² tumor characteristics, ³³ patterns of mRNA profiling and protein expression ^{34, 35} and genetic alterations. ³⁶ As well as being anatomically adjacent, GCA and esophageal squamous cell carcinoma(ESCC) occur at epidemic rates in this study population, share some etiological risk factors as well a GWAS risk variant in the PLCE1 gene. ¹³ We recently profiled gene expression levels in matched tissues from patients with GCA (n=41) and GNCA (n= 94) from this high-risk population. ³⁷ In agreement with previous studies we found a number of genes that were differentially expressed in GCA, but not GNCA, and vice versa. Added to this, differentially expressed genes reported in GCA were also dysregulated in a similar pattern in ESCC patients from this same population.³⁷ Collectively, this data may suggest etiological differences in the gastric carcinogenesis pathway, and in the exposures important for the development of GCA or GNCA in this high risk population. Differential roles for Fas signaling or specifically Fas-mediated apoptosis or proliferation may also be important in these gastric tumor subgroups. However, further studies are required to clarify the role of Fas-signaling in gastric carcinogenesis in cardiac versus non-cardiac tumors.

The strongest gene-based association observed for overall risk of GC (P = 0.0038) as well as risk of GNCA (P = 0.0127) in our study population was observed for MAP2K4, with a marginal non-significant association (P= 0.0529) for GCA. MAP2K1 was also significantly associated with risk of GC (P = 0.0185) and GCA (P = 0.0203) in our population, while MAPK8 was associated with GCA (P = 0.0436), GNCA (P = 0.0077), and GC risk overall (P = 0.0041). MAP kinase (MAPK)-related gene products frequently integrate signaling outputs of different signal transduction circuits including Fas-mediated apoptosis in a cell.^38-41MAP2K4, which encodes a map kinase kinase of JNK (JNKK1) and p38, is classically associated with growth arrest and apoptosis in cells and has been reported to be a metastasis suppressor involved in multiple cancer types. ^{38, 39}MAP2K1 encodes MEK1, which functions in the MAPK/ERK cascade. MEK1 can target peroxisome proliferator-activated receptor gamma (PPARG), a nuclear receptor that promotes differentiation and apoptosis, while activation of MEK1 in Jurkat T lymphocytes attenuates Fas-mediated apoptosis. ³⁹MAPK8 encodes the c_JUN N-terminal protein kinase JNK1, which is activated by JNKK1 (or the MAP2K4 product) and regulates the activity of c-Jun and c-Myc as well as the proapoptotic Bcl-2 family protein. ⁴¹ In addition, exonic genetic variation in MAPK has been observed in a majority of GC cell lines. ⁴²

The second strongest gene-based association observed with overall GC risk was for FAF1 (P =0.0039) an interaction partner of Fas, which was also significantly associated with risk of both GCA (P = 0.0265) and GNCA (P= 0.0412) in our population. Initially postulated to be a tumor suppressor, ²² FAF1 have functions in several biological processes including Fas-induced apoptosis, NF-κB signaling, ubiquitination, proteasomal degradation, canonical Wnt signaling and neuronal cell survival. ^{22, 43-45} We identified thirteen significant FAF1 SNPs (P<0.05) in strong LD (mean max r² = 0.96), representing three independent signals associated with reduced risk of GC. Given that FAF1 protein is an important mediator of apoptosis, it is plausible that one or more of these SNPs could alter expression of FAF1 or modify protein interactions that might alter apoptosis. Also, reduced FAF1 protein has been reported in a high percentage of human gastric carcinomas, most prominantly in carcinomas containing signet ring cells. ⁴⁶ A significant decrease in FAF1 mRNA expression was observed for Caucasian patients with cleft palate who were homozygous for the major T allele (TT genotype) for rs3827730 (P=0.0015). ⁴⁷ Although rs3827730 was not significant after correcting for multiple testing comparisons, the T allele of FAF1 rs3827730 was significantly associated with reduced risk of GC (per allele OR, 0.89, 95% CI, 0.80-0.99, P=0.026) and GCA (per allele OR: 0.88; 95% CI: 0.77-0.99; P = 0.039), but not GNCA, in the present study.

We also observed gene-based associations for CFLAR (P = 0.015), CASP10 (P= 0.011), and CASP8 (P= 0.013), which cluster on chromosome 2q32-q33, with overall risk of GC in our population. Furthermore, these genes were significantly associated with risk of GCA (CFLAR P= 0.020, CASP10, P = 0.015 and CASP8, P = 0.004), but not GNCA. CFLAR, CASP10 and CASP8 proteins regulate the extrinsic apoptosis pathway. CFLAR, which encodes the cellular FLICE-like inhibitory protein or c-FLIP, acts as an inhibitor of Fas-mediated apoptosis, ^{16, 17} and while bound to RIP2 can also mediate activation of NF-κB and/or non-apoptotic signals including cell proliferation. Both CASP8 and CASP10 are highly expressed (even co-expressed) in gastric adenocarcinomas, irrespective of histological subtypes and depth of invasion. ⁴⁸CFLAR mRNA and c-FLIP protein are also frequently elevated in gastric adenocarcinomas of Chinese patients. ⁴⁹ Using a meta-analysis of GWAS data from the study populations evaluated here and other population of Chinese ethnicity, we recently reported a strong association of five SNPs which map to 2q33 and the CASP8/ALS2CR12/TRAK2 gene region with risk of esophageal squamous cell carcinoma (ESCC). ⁵⁰ However, neither CASP8 rs10931936 (P = 0.8), which was included in the current study, nor the four remaining variants were shown to be associated with risk of GC in this population (data not shown), suggesting the latter association may be specific for ESCC.

IKBKB encodes a catalytically active-protein called IkappaB-kinase (IKKB) that is responsible (as part of a larger complex including IkappaA-kinase (IKKA)) for the dissociation of the inhibitor of NF-κB and its subsequent activation. ⁵¹ In this study, we observed a significant gene-based association for IKBKB with risk of GC (P=0.048) and GNCA (P = 0.048), but not GCA. CHUK, which encodes IKKA was not associated with risk of GC. IKBKB rs5029748 was identified as the most significant SNP in IKBKB in our study, and was associated with protection against GNCA (per allele OR: 0.86; 95%CI: 0.75-0.97, P= 0.018) as well as GC overall (per allele OR: 0.90; 95%CI: 0.82-0.98, P = 0.018). IKKB represents a key protein in the regulation of apoptosis in epithelial cells as well as in the reponse of gastrointestinal mucosa to external stimuli. ⁵¹ While the effects of loss of IKBKB on cancer risk appears to be tissue-specific, conditional knockout of IKBKB in the normal gastric epithelium of mice showed decreased mRNA expression of CFLAR, accelerated Helicobacter-dependent gastric apoptosis, proliferation, and the development of dysplasia. ⁵¹ However, little is known about the biological relevance of genetic variation in IKBKB and how this might influence the activity or protein interactions, as well as downstream NF-κB/IKKB-related processes such as apoptosis and inflammation.

Lastly, significant gene-based associations were observed for PAK2 and CASP8AP2 with risk of GC (P = 0.048 and P = 0.020, respectively), but not with risk of GCA or GNCA per se, a result which may reflect limited power. PAK2 encodes a Group 1 serine/threonine protein kinase (also called PAK2) and is the only member of the PAK family that is directly activated by CASP3, resulting in the morphological and biochemical changes of apoptosis. ⁵²CASP8AP2 encodes a pro-apoptotic protein called FLICE-Associated Huge (FLASH) that acts as a downstream mediator (together with FAF-1) in the activation of CASP8 in Fas-mediated apoptosis and NF-κB activation. ⁵³ Limited evidence indicates that somatic mutations in CASP8AP2 are rare in gastric carcinomas, but increased expression of FLASH has been detected in 70% of gastric carcinoma tissues compared to normal mucosa, suggesting that FLASH may play an important role in gastric carcinogenesis. ⁵³

In our study population, cases were more likely to use tobacco and to drink alcohol than controls, however these exposures are not major risk factors for GC in our Chinese populations ^{4, 10} and neither smoking nor alcohol drinking confounded our genotypic findings. This study had several strengths and several limitations. Our examination of a large number of SNPs associated with Fas signaling is a strength, in addition to our comprehensive assessment of both gene- and overall pathway associations. Examination of many SNPs does, however, create a concern over multiple testing. The large number of cases studied also allowed us to assess all of these risks with reasonable power. Despite the large size of our study, further studies are needed to replicate these findings. Another limitation of this study is that we were not able to examine SNP associations by H.pylori (Hp) status. Infection with H. pylori is prevalent in this high-risk region of north central China, presumably due to undeveloped living conditions. ⁵⁴ Thus, a very high prevalence of Hp-positive status in both cases and controls in this study limited our ability to evaluate this pathway in Hp-negative subjects. Finally, the generalizability of our findings to other ethnic populations remains to be determined.

In conclusion, our evidence suggests an important role for genetic variation in the Fas signaling pathway on risk of GC, and in particular GCA, in this high risk Chinese population. This association appears to be driven mainly by genetic variation in MAP2K4, FAF1, MAPK8, CASP10, CASP8, CFLAR, MAP2K1, CAP8AP2, PAK2 and IKBKB genes. Polymorphisms in these genes may result in altered expression, signaling, and/or interactions with other proteins that lead to changes in the apoptotic-proliferation phenotype and thus GC risk. Further investigation into the association of this pathway with risk of GC is warranted.

NOVELTY & IMPACT.

Although the incidence of gastric cancer (GC) is declining globally, it remains the 2^nd leading cause of cancer death worldwide, and it has a poor prognosis. Helicobacter pylori is acknowledged as the primary risk factor for GC. Evidence from genome-wide associations and other studies suggests genetics plays a role in the etiology of GC, particularly in high risk regions of the world such as China. Also, the deregulation of Fas signaling is a likely early and necessary alteration in the development of GC. Here we report a further analysis of data from a GC genome-wide association study conducted in ethnic Chinese. Specifically, we investigated the etiologic role of 53 genes in the Fas signaling pathway through a comprehensive evaluation of pathway-, gene- and SNP-based associations with GC, including both cardia and noncardia subsites. Results suggest an important role for genetic variation in the Fas signaling pathway on risk of GC, particularly cardia, in this high risk Chinese population. The identification of predisposing genetic factors associated with development of GC may ultimately lead to improved prognostic and therapeutic strategies.