Genetic variation in the retinoid X receptor and calcium-sensing receptor and risk of colorectal cancer in the Colon Cancer Family Registry

Abstract

Genetic variants in the calcium/vitamin D metabolic pathway may be related to risk for colorectal cancer. While several investigations of vitamin D receptor (VDR) polymorphisms and colorectal cancer have been conducted, no studies to date have evaluated the association of genetic variation in the heterodimer partner for VDR, the retinoid X receptor (RXR). Another important gene in this pathway is the calcium-sensing receptor (CASR). Employing a discordant-sibship case–control design, we examined the association between single nucleotide polymorphisms (SNPs) in RXRA and CASR and risk for colorectal cancer overall and by colorectal subsite and microsatellite instability (MSI) status using data from the Colon Cancer Family Registry. No gene-level relationships between RXRA or CASR and colorectal cancer overall were observed. However, for RXRA SNP rs7861779, a high-interest SNP selected for study a priori, there was a statistically significantly increased risk for proximal colorectal cancer among those with at least one A allele [odds ratio (OR) = 1.42; 95% confidence interval (CI) = 1.03–1.97]. Another selected RXRA SNP, rs12004589, was significantly associated with risk of MSI-high cancers (OR = 2.27; 95% CI = 1.13–4.56). Additionally, CASR SNP rs1801726 was significantly associated with a reduced risk for rectal cancer (OR = 0.53; 95% CI = 0.29–0.96). These results provide support that RXRA SNPs rs7861779 and rs12004589 and CASR SNP rs1801726 may be important markers for colorectal neoplasia. Further work is needed to elucidate their role in the carcinogenic pathway.

Introduction

Recently, the relationship between calcium, vitamin D, and colorectal cancer has garnered a great deal of attention (1–3), and genetic variation in metabolic pathways for these nutrients may play a role in colorectal carcinogenesis (3). Within the context of the Colon Cancer Family Registry, an investigation of two key genes in this pathway, the vitamin D receptor (VDR) and the vitamin D binding protein (GC), has recently been completed, and no main effects for these genes were reported (4). However, there are several other key genes in the calcium/vitamin D pathway that may be of importance, including the retinoid X receptor (RXR) and the calcium-sensing receptor (CASR) (3).

The RXR is part of the steroid nuclear receptor super-family, the members of which have been identified as potential targets for the prevention and treatment of several cancers (5–7); its ligands include 9-cis retinoic acid and an array of RXR-specific ligands called rexinoids (8). RXR can form homodimers or heterodimerize with other nuclear receptors, including the VDR and the peroxisome proliferator-activator receptor (PPAR) (9,10). Previous work from a study of colorectal adenoma demonstrated that genetic variation in RXRA, an isoform of RXR, was associated with the risk of colorectal adenoma recurrence (11); however, to our knowledge, there are currently no studies of polymorphisms in RXRA and risk for colorectal cancer.

Another gene of interest in this pathway is the CASR. In the promoter region of CASR, vitamin D response elements are present (12), providing support for mechanistic interactions between vitamin D and calcium. CASR may also independently have a role in colorectal carcinogenesis (3,13). Higher intake of calcium has been associated with reduced risk for colorectal neoplasms (2,14), and the CASR has a critical role in the maintenance of calcium homeostasis (15). Expression of this receptor has been shown to be reduced in colon cancer cells, and this decrease may result in the inability of these cells to utilize calcium in a protective manner (15). Genetic variation in CASR has been shown to be related to risk of colon cancer (16), particularly of the proximal colon (13), and advanced stage rectal cancer (17). However, to date, no studies have evaluated the association between genetic variation in RXR and CASR and colorectal cancer by microsatellite instability (MSI) status. The purpose of this study was therefore to comprehensively evaluate the relationship between genetic variation in RXRA and CASR and colorectal cancer within participants from the Colon Cancer Family Registry and to assess whether these relationships varied by colorectal subsite or by MSI status.

Materials and methods

Study population

Participants were men and women drawn from the Colon Cancer Family Registry, as described in detail elsewhere (18). The Colon Cancer Family Registry includes six recruitment centers: Cancer Care Ontario (Toronto, Canada), Fred Hutchinson Cancer Research Center (Seattle, WA, USA), Mayo Clinic (Rochester, MN, USA), University of Hawaii (Honolulu, HI, USA), University of Melbourne (Melbourne, Australia) and the University of Southern California Consortium (Los Angeles, CA, USA). The Colon Cancer Family Registry included both population-based and clinic-based recruitment, with the former conducted via cancer registries and the latter identified at high-risk cancer clinics (18). In the current work, we present data from the population-based participants only. Proband cases were enrolled in the study and subsequently asked to assist with recruitment of relatives into the registry as either additional cases or controls (18). Self-administered questionnaires or interviews were employed for ascertainment of such information as demographics, race and ethnicity, personal and familial history of cancer, medical history, reproductive history, physical activity and use of alcohol and tobacco (18). Informed consent from all study participants was obtained at each study center, and the protocol was approved at the individual Colon Cancer Family Registry sites.

Assessment of MSI

A total of 10 markers were employed for evaluation of MSI status for colorectal cancer cases (ACTC, BAT25, BAT26, BAT40, BAT34C4, D5S346, D17S250, D18S55, D10S197 and MYCL). Cases were classified as MS stable, MSI low and MSI high if 0% of loci were unstable, >0 to <30% of loci were unstable or ≥30% of loci were unstable, respectively (18,19). MSI status was assigned only in those cases with a valid result for at least four markers (18,19).

Genotyping

Genotyping was performed on the Illumina Golden Gate platform (Illumina®, San Diego, CA, USA). Briefly, DNA was activated with streptavidin/biotin and added to a hybridization mixture. After hybridization, the samples were washed, followed by extension, ligation and cleanup. Universal primers labeled with cyanine3 or cyanine5 were utilized for polymerase chain reaction labeling of the DNA. The labeled DNA was then allowed to hybridize with the Sentrix Array Matrix and placed in the BeadArray Reader for quantification of the fluorescence signal. Analysis of the output from the fluorescence reading was managed using Bead Studio software (Illumina®).

Single nucleotide polymorphism (SNP) selection was performed using Haploview Tagger to identify bin tags from a Caucasian European (CEU) population. Initial tag SNPs and linkage disequilibrium (LD) blocks were identified from HapMap data release #16c.1, June 2005, on NCBI B34 assembly, dbSNP b124. Tag SNPs from these data were identified utilizing the following criteria: minor allele frequency > 5%; pairwise r² > 0.95 and at least 60 bp distance between neighboring SNPs (20,21). SNPs located at the 5′ and 3′ ends of an LD block were also included. SNPs with little or no LD were selected from HapMap or dbSNP at a density of 1 per kb. SNPs were considered to have failed genotyping if they met at least one of the following: Illumina GenTrain Score < 0.4, 10% GC score < 0.25, AB T Dev > 0.1239, call frequency < 0.95, intra-plate replicate errors > 2, parent–parent–child errors > 2 or discordance with HapMap > 3. Participants were genotyped for a total of 41 SNPs in RXRA and 40 SNPs in CASR. Prior to the statistical analysis, all SNPs that failed the above criteria or were monomorphic were excluded from the data set (RXRA = 12; CASR = 4), leaving a total of 29 SNPs in RXRA and 36 SNPs in CASR for inclusion in the final analyses.

Statistical analyses

A discordant-sibship case–control study design was employed for the current work, which was restricted to the population-based participants. Any probands or their relatives who were diagnosed with invasive colorectal cancer were counted as cases, whereas unaffected siblings with no personal history of colorectal cancer were controls. All participants from the population-based Colon Cancer Family Registry who had available genetic data for SNPs in RXRA and CASR were included in the current analyses, yielding a total of 1802 cases and 2874 controls.

Principal components (PCs) analysis was employed to evaluate the overall association between variants in each gene and risk of colorectal cancer. First, PCs were computed as uncorrelated linear combinations of the original SNP data. These PCs are computed such that the first PC has maximal variance, the second maximizes the remaining variance and so forth. The PCs were modeled using conditional logistic regression with the unique sibship identifier as the matching factor, utilizing an 80% explained-variance threshold in determining how many PCs to include in the models (22). A P-value for the overall gene-outcome association was obtained from a likelihood ratio test comparing the model with PCs versus a model containing only age and sex, with degrees of freedom equal to the number of PCs. The data were further explored using PC analysis through stratification by cancer location (proximal, distal and rectal) and MSI status (low/stable and high). Additionally, a priori, we selected two high-interest SNPs in RXRA [rs7861779 (11) and rs12004589 (11)] and four high-interest SNPs in CASR [rs1042636 (13), rs12485716 (13), rs1801725 (16,17) and rs4678174 (13)] for analysis using a log-additive model. These SNPs were specifically chosen because they or an SNP with high LD were reported in the literature to have a statistically significant association with colorectal neoplasia (11,13,16,17). Single-SNP analyses were then performed on the remaining SNPs considering additive, recessive and dominant modes of inheritance. To ensure a correct type I error rate, we adjusted for the multiple correlated tests using the approach presented by Conneely and Boehnke (23). Variables examined for potential confounding included age, sex, body mass index, physical activity and baseline tumor site; however, none changed the model by 10% or greater (24). All models were therefore adjusted for age and sex only.

Results

Table I presents selected characteristics of the participants with genotype data in the Colon Cancer Family Registry. A total of 1747 population-based sibships were included in the current work. Data for MSI status were available for a total of 1182 of the cases, with 9.8% classified as MSI high, 8.4% as MS low, 47.5% as MS stable (Table I). There were no material differences in age, race, body mass index or physical activity between the cases and controls; however, a larger proportion of cases were men.

Table I.

Selected characteristics of cases and sibling controls from population-based participants in the Colon Cancer Family Registry

	Cases (n = 1802)	Sibling controls (n = 2874)
Person characteristic
Mean age ± SD	53.5 ± 10.9	54.0 ± 11.8
Sex, no. (%)
Male	924 (51.3)	1278 (44.5)
Female	878 (48.7)	1596 (55.5)
Race, no (%)
Non-hispanic white	1576 (87.5)	2507 (87.2)
Black	32 (1.8)	42 (1.5)
Asian	69 (3.8)	113 (3.9)
Othera	104 (5.8)	189 (6.6)
Unknown/missing	21 (1.2)	23 (0.8)
BMI (kg/m2)b
15–18 (underweight)	22 (1.2)	25 (0.9)
18–25 (normal)	628 (34.9)	1153 (40.1)
25–30 (overweight)	667 (37.0)	1035 (36.0)
30+ (obese)	422 (23.4)	592 (20.6)
Unknown/missing	63 (3.5)	69 (2.4)
Physical activity (MET hours)c
0–6 (inactive)	438 (24.3)	666 (23.2)
6.1–20 (less active)	491 (27.3)	776 (27.0)
20.1–44 (active)	412 (22.9)	641 (22.3)
44+ (very active)	377 (20.9)	631 (22.0)
Unknown/missing	84 (4.7)	160 (5.6)
Tumor characteristics
Site, no. (%)
Right colon	595 (33.0)	—
Left colon	525 (29.1)
Rectum	593 (32.9)
Unknown/missing	89 (4.9)
MSI, no. (%)d
MSS	855 (47.5)	—
MSI-L	151 (8.4)
MSI-H	176 (9.8)
Unknown/missing	620 (34.4)

MET, metabolic equivalent; MSS, microsatellite stable.

^aIncludes individuals who self-identified themselves as hispanic, native, Hawaiian/Pacific Islander and mixed race.

^bSelf-reported weight and height 2 years prior to questionnaire completion date used to calculate body mass index (BMI).

^cAverage weekly total lifetime MET hours.

^dNot all individuals were tested for MSI.

Results for PCs analyses conducted to evaluate whether there was an overall gene-level relationship between RXRA or CASR and colorectal cancer are presented in Table II, along with results for pre-selected RXRA and CASR SNPs. There were no significant gene-level associations between RXRA and CASR and colorectal cancer overall nor were there any single-SNP associations for either gene among SNPs selected a priori (Table II) or in an analysis of all single SNPs in each gene (data not shown).

Table II.

Association between RXRA and CASR and colorectal cancer in population-based participants in the Colon Cancer Family Registry

	ORa 95% CI
RXRA
PC1	1.00	0.97–1.04
PC2	1.00	0.92–1.08
PC3	1.00	0.92–1.10
PC4b	1.00	0.88–1.13
LRT P-valuec			1.00
CASR
PC1	1.00	0.95–1.05
PC2	0.99	0.93–1.05
PC3	0.96	0.89–1.02
PC4	1.08	1.00–1.17
PC5b	1.04	0.95–1.14
LRT P-valuec			0.21
RXRA SNPsd
rs7861779	0.98	0.82–1.18
rs12004589	0.99	0.82–1.21
CASR SNPsd
rs1042636	0.90	0.73–1.10
rs12485716	0.95	0.82–1.08
rs1801726	0.86	0.62–1.18
rs4678174	0.92	0.81–1.05

^aAdjusted for age and sex and matched for center and race.

^bAn 80% explained-variance threshold is used for including PCs in the model.

^cP-value for each model is from a likelihood ratio test (LRT) with degrees of freedom equal to the number of PCs.

^dORs were estimated assuming a log-additive model.

Table III presents findings for PCs analyses by colorectal subsite. No significant gene-level relationships for RXRA or CASR and proximal, distal or rectal cancers were observed. However, in analyses of single SNPs selected a priori, for RXRA SNP rs7861779, there was a statistically significant relationship with proximal colon cancer [odds ratio (OR) = 1.42; 95% confidence interval (CI) = 1.03–1.97; P = 0.03]. Additionally, CASR SNP rs1801726 was statistically significantly related to reduced risk for rectal cancer (OR = 0.53; 95% CI = 0.29–0.96). When conducting the analyses of all SNPs in CASR and after adjusting for multiple comparisons, one SNP (rs17203502) was statistically significantly related to proximal colorectal cancer in the recessive inheritance model (OR = 0.55; 95% CI = 0.40–0.78; supplementary Table I is available at Carcinogenesis Online), whereas no SNPs in RXRA were significantly associated at any colorectal subsite (supplementary Table II is available at Carcinogenesis Online). As shown in Table IV, there were no significant gene-level relationships between RXRA and CASR and risk for colorectal cancer by MSI status, though one of the RXRA SNPs selected a priori (rs12004589) was associated with MSI-high cancers (OR = 2.27; 95% CI = 1.13–4.56). No significant relationships for any of the pre-selected CASR SNPs were observed by tumor MSI status; further, after analyses of all available SNPs in RXRA and CASR and adjustment for multiple comparisons, no statistically significant associations were observed by MSI status (supplementary Tables III and IV are available at Carcinogenesis Online).

Table III.

Association between RXRA and CASR and colorectal cancer in population-based families in the Colon Cancer Family Registry

	Proximal colon cancer (cases = 595)			Distal colorectal cancer (cases = 525)			Rectal cancer (cases = 593)
	ORa	95% CI		ORa	95% CI		ORa	95% CI
RXRA
PC1	1.00	0.94–1.07		1.01	0.94–1.08		0.99	0.93–1.06
PC2	0.83	0.74–0.98		1.17	1.01–1.35		1.04	0.91–1.20
PC3	0.98	0.84–1.15		0.90	0.76–1.06		1.07	0.91–1.26
PC4b	1.07	0.87–1.32		0.94	0.84–1.35		0.86	0.68–1.10
LRT P-valuec			0.23			0.20			0.65
CASR
PC1	0.99	0.91–1.07		0.99	0.91–1.09		1.03	0.94–1.12
PC2	1.05	0.94–1.16		1.01	0.90–1.12		0.92	0.82–1.02
PC3	0.89	0.79–1.00		1.03	0.92–1.16		0.98	0.88–1.10
PC4	1.13	0.99–1.29		1.07	0.93–1.23		1.04	0.90–1.21
PC5b	1.12	0.96–1.31		1.06	0.90–1.25		0.96	0.82–1.11
LRT P-valuec			0.07			0.92			0.61
RXRA SNPsd
rs7861779	1.42	1.03–1.97		0.79	0.57–1.10		0.87	0.64–1.19
rs12004589	1.16	0.83–1.63		0.76	0.53–1.09		1.03	0.74–1.44
CASR SNPsd
rs1042636	0.83	0.58–1.19		0.90	0.62–1.30		0.97	0.68–1.39
rs12485716	1.01	0.80–1.29		0.93	0.72–1.20		0.93	0.73–1.18
rs1801726	0.86	0.49–1.51		1.21	0.69–2.14		0.53	0.29–0.96
rs4678174	0.99	0.78–1.25		0.97	0.76–1.23		0.84	0.67–1.06

^aAdjusted for age and sex and matched for center and race.

^bAn 80% explained-variance threshold is used for including PCs in the model.

^cP-value for each model is from a likelihood ratio test (LRT) with degrees of freedom equal to the number of PCs.

^dORs were estimated assuming a log-additive model.

Table IV.

Association between RXRA and CASR and colorectal cancer in the Colon Cancer Family Registry

	MSI low/stable (cases = 1006)			MSI high (cases = 176)
	ORa 95% CI			ORa 95% CI
RXRA
PC1	1.02	0.97–1.07		0.93	0.82–1.05
PC2	1.01	0.92–1.12		0.93	0.72–1.21
PC3	0.97	0.86–1.09		0.96	0.71–1.30
PC4b	1.04	0.88–1.24		1.22	0.84–1.78
LRT P-valuec			0.86			0.61
CASR
PC1	1.03	0.97–1.10		0.97	0.83–1.14
PC2	0.98	0.91–1.07		1.01	0.84–1.22
PC3	0.97	0.89–1.06		1.02	0.83–1.24
PC4	0.99	0.90–1.10		1.16	0.89–1.51
PC5b	1.03	0.91–1.15		0.90	0.67–1.21
LRT P-valuec			0.91			0.85
RXRA SNPsd
rs7861779	0.95	0.75–1.20		1.30	0.73–2.30
rs12004589	0.90	0.70–1.16		2.27	1.13–4.56
CASR SNPsd
rs1042636	0.82	0.63–1.07		1.27	0.66–2.43
rs12485716	1.03	0.86–1.23		0.86	0.58–1.30
rs1801726	0.91	0.60–1.39		1.20	0.32–4.52
rs4678174	1.01	0.85–1.21		0.88	0.59–1.32

^aAdjusted for age and sex and matched for center and race.

^bAn 80% explained-variance threshold is used for including PCs in the model.

^cP-value for each model is from a likelihood ratio test (LRT) with degrees of freedom equal to the number of PCs.

^dORs were estimated assuming a log-additive model.

Discussion

The results of the current work do not provide support for a gene-level relationship between RXRA and CASR and risk for colorectal cancer. However, the associations observed for one CASR SNP and two RXRA SNPs selected a priori are of interest. Carriage of at least one A allele at RXRA rs7861779 was shown to be associated with higher risk of proximal colorectal cancer, whereas the presence of the A allele at RXRA rs12004589 was linked with higher risk of MSI-high cancers. Additionally, the high-interest CASR SNP rs1801726 was significantly related to reduced risk of rectal cancer.

While the study of genetic variation in nuclear receptors has most often focused on the heterodimer partners of RXR such as VDR and PPAR, evidence supports a role for RXR itself in carcinogenesis. Ligands for RXRα, an isoform expressed in the colonic epithelium, enhance the interaction between RXR and β-catenin and induce β-catenin degradation (25,26). In combination with PPAR ligands, RXR agonists act synergistically to inhibit growth in colon cancer cell lines (27). RXR ligands also alter the cellular response to the VDR ligand 1,25(OH)₂D₃. In Caco-2 cells, the addition of RXR agonists increased the proliferative response, whereas in HT-29 cells, the response was blocked (28). Hence, experimental work indicates that RXR has an active role in carcinogenesis, and this role may vary by a complex network of potential heterodimer partner selections.

In previous work, carriage of at least one A allele for RXRA rs7861779 was shown to be associated with reduced odds of proximal colorectal adenoma (11), whereas the results of the current work show increased risk of proximal colorectal cancer with presence of at least one of these alleles. Any mechanism for explaining why RXRA would have different roles in adenoma as compared with cancer is purely speculative, particularly as rs7861779 is located within an intron. Although intronic SNPs can potentially affect alternative splicing of RNA (29–31), it is possible that this SNP is in LD with a site of genetic variation, which exerts a conformational change in the resultant protein, which in turn has biological effects. If this were the case, the functional change might be related to the capacity of RXRα to heterodimerize with specific nuclear receptor partners such as VDR and PPAR, both of which are of interest to colorectal carcinogenesis. The RXR/PPAR heterodimer is permissive and therefore can be activated by ligands for either receptor; provision of ligands simultaneously exerts a synergistic effect of cell growth inhibition and apoptosis induction above the activity of the ligands alone (32). If the structure of RXRα was changed such that it exhibited decreased affinity for PPAR heterodimerization and/or preference for an alternate receptor, this might lead to cancer progression. In turn, the same conformational change might be protective from the formation of new adenomas in the colorectum by increasing the affinity of RXR for VDR and thus activation of the latter, which may have a critical role in the maintenance of normal colon tissue after removal of adenomas (33,34). Overall, while the results of the present study were contrary to what was expected based on prior work involving adenomas, the repeated finding of an association for this particular SNP and proximal colorectal neoplasia indicates that it may be marking a region of RXRA that has the potential for activity in proximal colon carcinogenesis.

While it is unclear as to why RXRA rs7861779 has been found to be associated with proximal and not distal lesions, neoplasia of these colorectal subsites may represent biologically disparate pathologies with potentially distinct genetic etiologies (35). PPARγ, one isoform of an RXRA heterodimer partner, exhibits different patterns of expression in the proximal and distal colon and regulates different genes within each colorectal subsite (36). In addition, chromosomal instability is more often described in the distal region, and genetic instability and MSI are more often found in proximal neoplasia (37). Interestingly, an RXRA SNP in strong LD with rs7861779 and rs12004589 was significantly associated with MSI-high lesions in the current work. MSI is associated with defective mismatch repair (38), most often because of methylation of mutL humolog 1, a mismatch repair protein (19). Previous work has shown that RXR interacts with the T:G mismatch-specific thymine-DNA glycolylase, which has a critical role in initiating the repair of methylated DNA (39). Though the potential mechanism of action is currently unknown, this finding provides more evidence that these SNPs may be marking a functional change that affects colorectal neoplasia of a particular biological pathway, possibly involving MSI.

With regard to CASR, many epidemiological studies have reported that calcium intake is related to a reduced risk for colorectal neoplasia (2). In a randomized, placebo-controlled intervention trial, Baron et al. (14) demonstrated that daily supplementation with calcium carbonate significantly reduced the risk of adenoma recurrence compared with the placebo group. The CASR is critical for maintenance of circulating calcium concentrations. This receptor is present in colon epithelial tissue, but in colon cancer, this expression is decreased (15). The promoter region of CASR contains vitamin D response elements (12), supporting a mechanistic relationship between vitamin D and calcium, and genetic variation in CASR may increase or reduce calcium absorption and therefore have implications in colorectal carcinogenesis (3). There have been several prior studies of variation in this gene and risk for colorectal cancers. In work by Dong et al. (13), the authors reported that whereas there were no associations with colorectal cancer overall, several CASR SNPs were significantly associated with proximal lesions (rs10934578, rs12485716, rs4678174 and rs2270916); while in the current work, these SNPs were not related to colorectal cancer. Another high-interest SNP in CASR is rs1801725; two studies reported a significant link with colorectal cancer (16,17), whereas others found no relationship (13,40,41). Specifically, Speer et al. (17) reported an association between this SNP and more advanced rectal tumors, and the results of the current work also reveal a significant association for this SNP that was observed only with rectal cancers. When conducting analyses of all CASR SNPs and adjusting for multiple comparisons, rs17203502 was significantly associated with risk for proximal colorectal cancer for the recessive mode of inheritance (supplementary Table I is available at Carcinogenesis Online). In another study of colorectal cancer, this intronic SNP was not associated with any cancer or proximal lesions; however, in that work, only one mode of inheritance was investigated (13).

Whereas there are several published reports for CASR and colorectal neoplasia, to our knowledge only one study, discussed above (11), has evaluated association of RXRA and colorectal lesions. However, there are two published papers investigating RXRA and other cancer sites. In one study, genetic variation in RXRA and risk for prostate cancer was investigated and no relationship was observed (42), though only 11 SNPs in RXRA were included in the work. In another study, no significant associations between two RXRA SNPs (rs1536475 and rs1805343) and biliary tract cancers were detected (43); these SNPs were not significantly related to colorectal cancer risk in the current work, nor with metachronous colorectal adenoma in another study (11).

A limitation of this work includes the lack of statistical power for investigating racial or ethnic differences in the reported associations. Because of relatively low numbers of non-white study participants, we could not conduct robust stratified analyses. Given the strong relationship between vitamin D and race/ethnicity, replication of these findings in a more diverse population is necessary.

In summary, the current work provides support for a role of CASR and RXRA in the development of colorectal cancer. The results for CASR support prior findings of a possible role in rectal cancers specifically, whereas those for RXRA provide further evidence of an important functional role in the proximal colon, which may be related in part to tumor MSI status.

Supplementary material

Supplementary Tables I–IV can be found at http://carcin.oxfordjournals.org/

Funding

National Cancer Institute, National Institutes of Health under RFA # CA-95-011 and through cooperative agreements with the Australasian Colorectal Cancer Family Registry (U01 CA097735); University of Southern California Familial Colorectal Neoplasia Collaborative Group (U01 CA074799); Mayo Clinic Cooperative Family Registry for Colon Cancer Studies (U01 CA074800); Ontario Registry for Studies of Familial Colorectal Cancer (U01 CA074783); Seattle Colorectal Cancer Family Registry (U01 CA074794); University of Hawaii Colorectal Cancer Family Registry (U01 CA074806); National Cancer Institute grants (CA-41108, CA-23074, CA-77145); Specialized Program of Research Excellence in Gastrointestinal Cancer (CA95060); Career Development Awards from the National Cancer Institute (1K07CA10629-01A1 to E.T.J., T32 CA009142 to J.N.P.).

Glossary

Abbreviations

CASR

calcium-sensing receptor

CI

confidence interval

LD

linkage disequilibrium

MSI

microsatellite instability

OR

odds ratio

PC

principal component

PPAR

peroxisome proliferator-activator receptor

RXR

retinoid X receptor

SNP

single nucleotide polymorphism

VDR

vitamin D receptor