Phospholipase A2G1B polymorphisms and risk of colorectal neoplasia

Abstract

Pancreatic phospholipase A₂, product of PLA2G1B, catalyzes the release of fatty acids from dietary phospholipids.Diet is the ultimate source of arachidonic acid in cellular phospholipids, precursor of eicosanoid signaling molecules, linked to inflammation, cell proliferation and colorectal carcinogenesis. We evaluated the association of PLA2G1B tagging single-nucleotide polymorphisms with colorectal neoplasia risk. A linkage-disequilibrium-based tagSNP algorithm (r²=0.90, MAF≥4%) identified three tagSNPs. The SNPs were genotyped on the Illumina platform in three population-based, case-control studies: colon cancer (1424 cases/1780 controls); rectal cancer (583/775); colorectal adenomas (485/578). Evaluating gene-wide associations, principal-component and haplotype analysis were conducted, individual SNPs were evaluated by logistic regression. Two PLA2G1B variants were statistically significantly associated with reduced risk of rectal cancer (rs5637, 3702 G>A Ser98Ser, p-trend=0.03; rs9657930, 1593 C>T, p-trend=0.01); principal component analysis showed that genetic variation in the gene overall was statistically significantly associated with rectal cancer (p=0.02). NSAID users with the rs2070873 variant had a reduced rectal cancer risk (P-inter=0.02). Specific associations were observed with tumor subtypes (TP53/KRAS). The results suggest that genetic polymorphisms in PLA2G1B affect susceptibility to rectal cancer.

Introduction

Colorectal cancer (CRC) is the third most common cancer worldwide [1]. Substantial advances have been made in understanding the molecular mechanisms of colorectal carcinogenesis, including the role of inflammation in cancer development and progression [2,3].

Eicosanoids derived from arachidonic acid (AA) have important roles in inflammation and a wide range of other pathophysiologic processes[4,5]. Laboratory and epidemiologic studies link aberrant arachidonic-acid metabolism, via the production of prostanoids (COX pathway) or leukotrienes (by lipoxygeneases), to the promotion of carcinogenesis [6-9]. The use of non-steroidal-anti-inflammatory drugs (NSAIDs), which inhibit PTGS (COX)-mediated conversion of AA to prostanoids is associated with decreased risk of CRC [10,11] and PTGS2 over-expression is directly correlated with the degree of dysplasia [12]. NSAID effects on survival may be mediated by PIK3CA mutation status [13].

Eicosanoid synthesis utilizes AA released from membrane phospholipids by specialized PLA2s [4] but the ultimate source of AA is diet. Mammals cannot synthesize AA de novo and must obtain the fatty acid or a precursor from dietary lipids [14]. During digestion, pancreatic phospholipase A₂ (PLA2G1B), secreted into the intestinal lumen and proteolytically activated, releases fatty acids from the sn-2 position of dietary phospholipids [15-17]. This fosters incorporation of AA into membrane phospholipids of cells throughout the body, available for subsequent release and use in eicosanoid signaling.

Given the role of PLA2G1B in supplying fatty acids as precursors for eicosanoid synthesis and in releasing lysophospholipid signaling molecules [18], we hypothesized that genetic variation in PLA2G1B may affect colorectal neoplasia. We evaluated the association of PLA2G1B tagSNPs with colorectal adenoma, colon cancer, and rectal cancer. We further examined whether the reduction of CRC risk by use of NSAIDs differs by specific haplotypes and genotypes [18]. The availability of tumor samples allowed us to analyze for differences in mutations of TP53 and K-ras [19].

Material and methods

The analyses are based on three US population-based case-control studies of colorectal adenomas [20], colon cancer [21], and rectal cancer [22] using subjects with available DNA from blood and tissue samples. Methods have been described in detail elsewhere [20-22]. Participants consented and the Institutional Review Board at FHCRC approved the study.

Colorectal adenoma cases (n=485) and polyp-free controls (n=578) were recruited through a large, multiclinic, gastroenterology practice in the Twin Cities area of Minnesota. Eligible participants: were aged 30-74 years; first diagnosed with a colorectal adenoma between 1991-1994; had no known genetic CRC syndrome; had and no history of cancer (except non-melanoma skin cancer), prior colorectal polyps, or inflammatory bowel-disease. All participants underwent colonoscopy; participation was 68%.

Colon cancer cases (n=1424) and controls (n=1780) and rectal cancer cases (n=583) and controls (n=775) were recruited from Utah, the Northern California Kaiser Permanente Medical Care Program (KPMCP), and the Twin Cities Metropolitan area of Minnesota (colon only). Participants aged between 30-79 years with no previous diagnosis of CRC, familial adenomatous polyposis, Crohn’s disease or ulcerative colitis were eligible. Colon cancer cases were first diagnosed 1991-1994 [21] whereas rectal cancer cases – including cancer of the rectosigmoid junction, or rectum – were first diagnosed 1997-2001 [22]. Participation among contacted colon cancer cases was 76% (controls: 69%), among contacted rectal cancer cases 73% (controls: 69%).

Information on health behaviors, anthropometry, medical history, family history of cancer, medication, and demographics were obtained by questionnaire as described previously (referent year 2 years prior to diagnosis/selection) [20-22]. A history of regular use of NSAIDs was defined as using any NSAID at least twice/week for ≥1 month.

Tumor DNA was obtained from paraffin-embedded tissue, categorized by TP53 or KRAS mutations, microsatellite instability (MSI) or the CpG-island methylator phenotype (CIMP) as previously described [23-26]. The proportion of MSI+ tumors in rectal cases was <3% and thus not investigated further. To compare cancer patients with specific molecular types of tumors controls, a generalized-estimating equation with a multinomial outcome was used.

We applied a linkage-disequilibrium (LD)-based, tagging-single-nucleotide-polymorphism (tagSNP) selection algorithm (r²≥0.90, MAF≥4%) to identify 3 tagSNPs in PLA2G1B (rs5637 3702 G>A Ser98Ser, rs9657930 1593 C>T and rs2070873 3027 G>T). Germline DNA was extracted from buffy coats for genotyping.

We used the same genotyping platform (Illumina^TM GoldenGate) for all three studies. Intraplate and interplate replicates and blinded duplicates were included (5%) for quality control, as were data from 30 CEPH trios (Coriell Cell Repository) genotyped by HapMap. Genotypes were excluded if any of the following was true: GenTrain Score <0.4; 10% GC Score <0.25; AB T Dev >0.1239; Call Frequency <0.85; Replicate Errors >2; P-P-C Errors >2; <85% concordance with blinded or non-blinded duplicates; or Hardy-Weinberg p-value >0.05.

Unconditional logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for associations between genotypes and outcomes. Genotypes were modeled using indicator variables for the heterozygous and the homozygous variant genotypes (unrestricted/co-dominant model); the dominant model (combining heterozygous and homozygous variants) was used if <10 cases or controls were involved. Models were adjusted for age, sex, and study center. For trend tests, genotypes were treated as a continuous variable. Analyses were restricted to non-Hispanic Caucasians (97% in the adenoma study, 91% and 82% in the colon and rectal cancer studies). A two-sided p-value <0.05 was considered statistically significant.

For principal component analysis (PCA) [27], we determined the number of principal components that explained >80% of the variance in the gene and performed logistic regression using these components. Gene-level significance was determined using a likelihood-ratio test. Each PCA was adjusted for age, sex, and study center.

Effect modification of the genetic association by NSAID use was evaluated by testing for a difference in trends within strata (never vs. ever) of NSAID use. Because use of NSAIDs may be confounded by other risk factors for colorectal neoplasia, we adjusted these analyses for age, sex, study center, smoking status, body mass index, physical activity, and intakes of calcium, total energy, and dietary fiber.

Results

Characteristics of the three study populations are presented in Table 1. Adenoma patients were younger than colon and rectal cancer patients. Overall tumors in the CRC cases (n=2007) were distributed approximately equally in the rectum, distal colon, and proximal colon. The tagSNPs in PLA2G1B were in modest LD (r²<0.6).

Table 1.

Characteristics of the three study populations^a

	Adenoma Study			Colon Cancer Study			Rectal Cancer Study
	Cases (N=485)	Controls (N=578)	p-value	Cases (N=1424)	Controls (N=1780)	p-value	Cases (N=583)	Controls (N=775)	p-value
Age	Mean (SD)	Mean (SD)		Mean (SD)	Mean (SD)		Mean (SD)	Mean (SD)
	58 (9.6)	52.9 (11.0)	<0.01	65.2 (9.7)	65.1 (10.3)	NAb	62.3 (10.8)	62.6 (10.5)	NAb
Location	N (%)	N (%)		N (%)	N (%)		N (%)	N (%)
Proximal	104 (22)	NA	NA	688 (50)	NA	NA	NA	NA	NA
Distal	300 (62)	NA		700 (50)	NA		NA	NA
Rectal	77 (16)	NA		NA	NA		583	NA
Sex
Male	304 (63)	227 (39)	<0.01	797 (56)	946 (53)	NAb	346 (59)	428 (55)	NAb
Female	181 (37)	351 (61)		627 (44)	834 (47)		237 (41)	347 (45)
Study Center
Kaiser Northern California	NA	NA	NA	617 (43)	647 (36)	<0.01	349 (60)	449 (58)	0.48
Minnesota	485 (100)	578 (100)		565 (40)	791 (44)		NA	NA
Utah	NA	NA		242(17)	342 (19)		234 (40)	326 (40)
Regular use of aspirin or NSAIDs
Yes	180 (37.1)	257 (44.5)	0.02	562 (39.5)	865 (48.6)	<0.01	263 (45.1)	417 (53.8)	<0.01
No	305 (62.9)	321 (55.6)		862 (60.5)	915 (51.4)		320 (54.9)	358 (46.2)
Body mass index
Normal/Underweight	159 (33.5)	225 (39.8)	0.10	475 (33.5)	708 (39.8)	<0.01	184 (31.7)	258 (33.5)	0.31
Overweight (25-29.9)	204 (43.0)	213 (37.7)		578 (40.7)	726 (40.9)		242 (41.7)	325 (42.2)
Obese (30+)	111 (23.4)	127 (22.5)		366 (25.8)	343 (19.3)		155 (26.7)	187 (24.3)

^aPercentages may not total to 100% due to rounding and missing values.

^bNA – these were matching factors.

Polymorphisms in PLA2G1B were statistically significantly associated with rectal cancer risk. PCA showed statistically significant results for the gene-level association with rectal cancer (p=0.02) (Table 2). For two PLA2G1B variants, rs5637 (3702 G>A Ser98Ser) and rs9657930 (1593 C>T), we observed statistically significant trends (p-trend=0.03 and 0.01) towards reduced risk of rectal cancer, with a greater number of variant alleles (Table 2). Carrying two variant alleles of rs5637 or rs9657930 in PLA2G1B was associated with >50% reduction in the risk of rectal cancer for the homozygous variant compared to the wild-type genotype in a log additive model (rs5637, p-trend=0.03; rs9657930, p-trend=0.01). The haplotype ACC was statistically significantly associated with lower rectal cancer risk (OR: 0.70; 95% CI (0.54-0.92), (Table 3). We did not observe any statistically significant associations for colorectal adenomas or colon cancers.

Table 2.

TagSNPs in PLA2G1B and risk of colorectal neoplasia, adjusted for age, sex and study center

		Adenomaa					Colon Cancerb					Rectal Cancerb
SNP	Genotype	Cases	Controls	OR	95% Cl	p-trendd	Cases	Controls	OR	95% CI	p-trendd	Cases	Controls	OR	95% CI	p-trendd
rs5637	GG	335	414	1.00	.		1022	1284	1.00	.	.	428	536	1.00		.
3702 G>A	GA	134	158	1.05	0.79-1.40		362	461	0.99	0.84-1.16	.	147	215	0.85	0.66-1.08	.
Ser98Ser	AA	14	10	1.67	0.69-4.04	0.40	36	27	1.70	1.02-2.82	0.39	8	23	0.43	0.19-0.98	0.03
rs2070873	GG	366	447	1.00	.	.	1082	1329	1.00	.	.	436	574	1.00	.	.
3027 G>TGT	109	129	1.09		0.80-1.48	.	307	405	0.93	0.79-1.10	.	139	181	1.00	0.78-1.29	.
	TT	7	6	1.34	0.43-4.20	0.49	24	34	0.85	0.50-1.44	0.32	6	19	0.42	0.17-1.06	0.36
rs9657930	CC	388	476	1.00	.	.	1161	1446	1.00	.	.	488	614	1.00		.
1593 C>T	CT	87	103	1.08	0.77-1.50	.	234	311	0.94	0.78-1.13	.	89	148	0.75	0.56-1.00	.
	TT	7	2	3.47	0.70-17.29	0.30	14	13	1.34	1.63-2.87	0.76	3	10	0.39	0.11-1.43	0.01
		Principal component analysisc p=0.49					Principal component analysisc p=0.60					Principal component analysisc p=0.02

^aAdjusted for age and sex.

^bAdjusted for age, sex, and study center.

^cPrincipal component analysis (PCA) tests for statistical significance at the gene level overall.

^dTests for significant trends with an increasing number of alleles.

Table 3.

PLA2G1B haplotypes (≥5% frequency in controls) and risk of rectal cancer

Haplotypea	Case (%)	Control (%)	OR (95% CI)
GGA	73.0%	69.0%	1.00 (ref.)
GTA	13.0%	14.1%	0.86 (0.68-1.08)
AGC	8.2%	10.8%	0.70 (0.54-0.92)
ATA	5.8%	6.0%	0.90 (0.65-1.24)
global p=0.05

^aOrder of SNPs in haplotype: rs5637 (G>A), rs2070873 (G>T), rs9657930 (C>T).

Associations stratified by molecular subtypes of colon cancer are shown in Supplementary Table 1. Homozygous variant genotypes were significantly associated with colon tumors characterized by TP53 or KRAS mutations (rs5637 and rs9657930) even though this analysis should be considered exploratory, because the number of cases was small. No association was observed for CIMP+ or MSI+ colon cancers,Supplementary Table 2.

For rectal cancer, we observed a statistically significant interaction between NSAID use and rs2070873, Supplementary Table 3. Among those with the wild-type GG genotype, NSAID use was associated with a ~40% reduction in risk. The variant genotype was associated with a ~30% reduction in risk, but no additional reduction in risk was observed with use of NSAIDs. PLA2G1B genotypes did not influence survival (data not shown).

Discussion

Our results provide the first evidence that genetic polymorphisms in the PLA2G1B gene can modify susceptibility to rectal cancer. Individuals with variant alleles for rs5637 and rs9657930 had a >50% reduced risk. It was proposed, almost 30 years ago, that CRCs distal to the splenic flexure involve different etiologic factors from proximal cancers [28,29]. The data presented here support this notion. Most recently it has been argued that a continuum of molecular pathologic patterns, rather than a dichotomy, exists across lower gastrointestinal cancers [30,31].

We attempted replication of our main effects in the GECCO and CCFR consortia (for consortia description see [32,33]). Our strongest hit, SNPs rs9657930, was associated with lower rectal cancer risk in the larger US study populations (i.e. CCFR and WHI); however, discordant results were observed in the GECCO-consortium’s European-based studies DACHS and ASTERISK, as well as some other US-based studies with small number of rectal cases, resulting in inconsistent replication (Figure S1). Underlying differences in lifestyle, BMI, and NSAID use could be causes for this inconsistency. No molecular tumor subtypes are yet available from this consortium.

Although we observed an association between SNPs in PLA2G1B with rectal cancer risk, we found no association with risk of the precursor lesions, i.e., rectal adenomas; this may have been due to the limited statistical power. Nevertheless, it remains possible that genetic variability in these PLA2G1B SNPs affects the progression from adenoma to cancer, rather than the early stages of neoplasia.

Our results indicate that general NSAID use is associated with a reduced risk of rectal cancer, as shown previously [34]. An important new finding here, however, is that the risk reduction may be limited to individuals with the PLA2G1B rs2070873 wild-type genotype, further supporting the concept of NSAID pharmacogenetics [34,35].

The SNPs identified as risk modifiers of rectal cancer most probably differ in their impact on the function of the PLA2G1B protein. TagSNP rs5637 is in exon 1, but is a synonymous mutation. Therefore, rs5637 could plausibly impact mRNA stability or protein expression levels and lead to altered PLA2G1B activity. PLA2G1B activity is a major determinant of digestive intake of fatty acids from dietary phospholipid. Thus, PLA2G1B SNPs could affect the supply of arachidonic acid available for loading into cellular phospholipids thereby modulating eicosanoid signaling in all tissues [36,37]. As noted, arachidonic acid is metabolized by the PTGS (COX)/LOX pathways to prostaglandins and leukotrienes, which have been shown to influence carcinogenesis and specifically colon cancer risk [6-9]. It may be relevant that PLA2G1B rs5637 was previously found to be associated with an increased risk of obesity in women [38]: obesity is a well-defined risk factor for CRC [39]. The PLA2G1B tagSNP rs9657930 and SNP rs2070873 are located in introns of the gene and could have indirect effects on PLA2G1B activity via altered mRNA processing or splicing.

Eicosanoids may modulate tumor progression through several mechanisms: e.g., by activating receptors on tumor epithelial cells to regulate cell proliferation, apoptosis, and migration/invasion or by inducing epithelial cells to secrete growth factors, pro-inflammatory mediators, and angiogenic factors that switch the microenvironment to one that supports tumor growth [40]. The eicosanoid prostaglandin E₂ mediates pro-inflammatory and tumor-promoting effects of PTGS2, and is strongly connected to CRC development and progression [41].

We observed statistically significant results with colon, but not rectal, tumors after stratification on TP53 and KRAS. Because of the limited sample size for these subset analyses, results should be considered suggestive. Combining research efforts on interactions of molecular changes and i.e. clinical outcomes could possibly lead to a better understanding, described recently as “molecular pathologic epidemiology” by Ogino et al [42-45].

The study has several strengths. The design, including adenoma patients, as well as those with both colon and rectal cancer, covers the range of the colorectal carcinogenesis paradigm. Comparable results are ensured by the use of standardized methods. Molecular subtyping in the colon and rectal cancer studies allows exploration of subtype-specific associations. One limitation of the study is the lack of information on functional impact of the identified SNPs. In this regard, we have used in silico approaches as first steps. Further observational and functional follow-up studies are needed for rs5637 (associated with obesity) and rs9657930.

In conclusion, the results provide evidence that polymorphisms in the PLA2G1B gene contribute to the risk of rectal cancer and may be associated with colon tumor subtypes. Protection by NSAID use against rectal cancer was confirmed, but may be limited to individuals with specific PLA2G1B genotypes.

Disclosure of conflict of interest

None declared.