Supplementary Results
Sensitivity analysis
There was no obvious bias between rs5275 genotype and diagnosis of familial CRC toward either the Australian or Spanish populations in this study. When the smaller set of pedigrees of Spanish origin were removed, the dichotomous and age of diagnosis phenotype p-values remained relatively stable (p <= 0.05). One notable exception was the heterozygous advantage model under the dichotomous phenotype, where removal of the Spanish pedigrees reduced support for the model (p = 0.1013). With further testing of the association using PBAT, we noted the PBAT result was particularly sensitive to using the more computationally expensive algorithm, which completely models extended pedigrees. A new rapid algorithm, which reduces the extended pedigrees into clustered sets of trios, resulted in no significant association of rs5275 with the dichotomous phenotype (Supplementary Table 3). However, the association was still present in the time-to-diagnosis analysis under the heterozygous advantage model (FBAT-Wilcoxon, p = 0.0089).
We wished to further test sensitivity by limiting the analysis to only considering nuclear family information. FBAT, the predecessor to PBAT, breaks all extended pedigrees into nuclear families and evaluates their contributions independently. As such, it cannot make use of the avuncular or cousin affected pair information in the pedigrees and the analysis has some similarity to the trio-based PBAT algorithm. Using FBAT with the dichotomous phenotype, there is no evidence to reject the null hypothesis (Supplementary Table 4). We observe the p-values generated by FBAT were similar to those of the trio-based PBAT algorithm (Supplementary Table 3). However, only PBAT could test under a heterozygous-advantage model.
It is also possible to form an association statistic using discordant relative pairs instead of affected pairs. The GDT extends the transmission disequilibrium test to consider the difference in number of alleles between discordant relative pairs. However, with a complex disease phenotype such as cancer, there is an expectation of incomplete penetrance - especially in younger individuals. While this reduces the sensitivity of such a test, it is possible to form many discordant relative pairs across the 147 pedigrees used in this study. The software created 346 discordant relative pairs and under the GDT association test there was no evidence to reject the null hypothesis (p = 0.65). Since many unaffected people in the study were relatively young, we also conducted an analysis with unaffected people less than 55 years of age recoded with an unknown affection status. A GDT association test of the resulting 212 relative pairs was also non-significant (p = 0.63).
Supplementary Discussion
Sensitivity analysis
The lack of support for a significant association using the FBAT software suggests the analysis is particularly sensitive to using an approach that makes full use of the extended pedigree structure. This sensitivity is also apparent with the PBAT trio-based algorithm, as a significant association was found in only one instance – with age of diagnosis under a heterozygous advantage genetic model. While the GDT does make full use of extended pedigrees, unlike FBAT and PBAT, it considers dichotomous relative pairs instead of affected relative pairs. Given the complex phenotype, likely incomplete penetrance and inclusion of relatively young ‘unaffected’ people in this study, we suggest these are not optimal conditions for power using a GDT test. Further, by considering the sum of allele counts in cases-control relative pairs, the GDT software is not designed to detect associations with an apparent heterozygous advantage genetic model.
Comparison of results with other cancers
Extrapolation of CRC associations with rs5275 genotype to other cancers is difficult. The relationship between rs5275 genotype and apparent cancer risk is dependent upon the cancer tissue of origin. Also, environmental influences, gender bias and ethnicity further complicate comparisons. For example, we note that most studies reporting an association between rs5275 and lung cancer were conducted in Asian populations with cohorts containing 68 – 73% men and many self-reported smokers. Unlike in CRC, the prevailing evidence suggests the C allele is protective in bladder1, breast2, lung3-5, oral premalignancies6 and ovarian7 cancers. Conversely, there is evidence to suggest the CC genotype poses risk for basal cell carcinoma8, gastric9 and prostate cancers10,11. Collectively, these disparate associations are not likely due to false positive associations, as the allelic association grouped by tissue of origin is often consistent. We note that a recent stratified meta-analysis of 23 studies of rs5275 found that the CC variant allele was associated with a significantly decreased risk of breast cancer compared with TT (OR = 0.84, p = 0.05) or TC+TT (OR = 0.83, p = 0.03), while the heterozygous CT allele was a risk to carriers in an other cancers group, which included colorectal, gastric, bile duct and basal cell cancers12. The protection by the CC genotype in breast and ovarian cancer may relate to the protection observed in younger women in this present study. Some studies of the association of rs5275 genotype with other cancers have stratified the data and from these we may ask if factors such as gender contribute to the association. A stratified analysis of bladder cancer has shown the association was only robust in males1. A recent gastric cancer study of 1681 cases (73.7% male) and 1916 controls found the C allele only posed risk for females (OR = 1.42)9. An association between rs5275 and gender was also observed in a recent study of Chinese with lung cancer5. Interestingly, while another two PTGS2 and two CYP2E1 SNPs were also examined, only rs5275 had a risk stratified by gender. In conclusion, we speculate that amongst PTGS2 SNPs, gender discordant associations may be a particular phenomenon of rs5275 but not limited to CRC.
Supplementary Table 1. VariantSEQr™ (Applied Biosystems) primers used to resequence PTGS2
Probe Accession / Probe Name / Forward Primer / Reverse PrimerPr001043787.1 / RSA001309903 / tgtaaaacgacggccagtTGTGGCTGAACAAATTAACGAAGCA / caggaaacagctatgaccTGGATGCAAAGAGGCTAGTGCC
Pr001044140.1 / RSA001308942 / tgtaaaacgacggccagtGCCAGGCTTGATTCCAATGC / caggaaacagctatgaccAAATGCCAAATTTATTAAGGTGGTGGA
Pr001044142.1 / RSA001308944 / tgtaaaacgacggccagtTCCACCACCTTAATAAATTTGGCATTT / caggaaacagctatgaccCGATGTTTCCAATGCATCTTCCA
Pr001044144.1 / RSA001308946 / tgtaaaacgacggccagtCACAAGTATGACTCCTTTCTCCGCA / caggaaacagctatgaccTGTCTTCATCGCCTTCACAGGA
Pr001044154.1 / RSA001308959 / tgtaaaacgacggccagtGCGCTTCCGAGAGCCAGTTC / caggaaacagctatgaccCCTGCAAATTCTGGCCATCG
Pr001044421.1 / RSA001308572 / tgtaaaacgacggccagtACACAACCCAAATTCCCAGGTTT / caggaaacagctatgaccGCCTATGTGCTAGCCCACAAAGAA
Pr001044772.1 / RSA001272351 / tgtaaaacgacggccagtTTATCCATGCGTGCGACGTG / caggaaacagctatgaccTTCTAGGCTGGTGTCCCATTGAA
Pr001045626.1 / RSA001262668 / tgtaaaacgacggccagtTTCCAACACAGTGTCGCAGTGAA / caggaaacagctatgaccAAACCTGGGAATTTGGGTTGTGT
Pr001069860.1 / RSA000609390 / tgtaaaacgacggccagtAATCCGGGCTTTCCTGGGAG / caggaaacagctatgaccCTCCTCAGCAGCGCCTCCTT
Pr001071083.1 / RSA000567952 / tgtaaaacgacggccagtGAAAGGGCATTAATTAGAATGGGAACG / caggaaacagctatgaccGTGCATTGGAATCAAGCCTGG
Pr001071084.1 / RSA000567954 / tgtaaaacgacggccagtAGCCTCTTTGCATCCATCTTGG / caggaaacagctatgaccTGCACTGCAGGCCTGGTACTC
Pr001071201.1 / RSA000564586 / tgtaaaacgacggccagtCTGGACGTGCTCCTGACGCT / caggaaacagctatgaccCAGCCTTTCTTAACCTTACTCGCCC
Pr001071208.1 / RSA000564530 / tgtaaaacgacggccagtGCTTGGCTTCCAGTAGGCAGGA / caggaaacagctatgaccCCTAGTGATCCGCCGGCTTC
Pr001071213.1 / RSA000564550 / tgtaaaacgacggccagtCAGCAATTTGCCTGGTGAATGA / caggaaacagctatgaccGCAGTTGTTCCAGACAAGCAGGC
Pr001071215.1 / RSA000564552 / tgtaaaacgacggccagtGGGTGTTAAATTCAGCAGCAATACGA / caggaaacagctatgaccTCCCTGAGCATCTACGGTTTGC
Pr001075665.1 / RSA000347162 / tgtaaaacgacggccagtCAACTTCTCTGTTTCCATTTGACCC / caggaaacagctatgaccAGGTTGCTGGTGGTAGGAATGTT
Pr001075667.1 / RSA000347165 / tgtaaaacgacggccagtAAATTCAATGGGACACCAGCC / caggaaacagctatgaccAACACCCTCTATCACTGGCATCC
Pr001075668.1 / RSA000347171 / tgtaaaacgacggccagtAGCCCGTTGGTGAAAGCTGG / caggaaacagctatgaccGCAAATGAGCGTCTTGGTATAATGTC
Pr001075669.1 / RSA000347175 / tgtaaaacgacggccagtCCAAAGGACAAACTTACGTGTTGAG / caggaaacagctatgaccCCAGGAAAGCCCGGATTATG
Pr001078664.1 / RSA000032321 / tgtaaaacgacggccagtGCATCATGGAAGATGCATTGGA / caggaaacagctatgaccTTCAAATCATCAACACTGCCTCAA
Pr001078666.1 / RSA000032335 / tgtaaaacgacggccagtTCTGTTGTGTTCCCGCAGCC / caggaaacagctatgaccTCAGTTTGTAGCTTTGGTGGATAAACA
Pr001078667.1 / RSA000032338 / tgtaaaacgacggccagtGACATTATACCAAGACGCTCATTTGC / caggaaacagctatgaccGGATAGTGAGCAGATGGCTACCTGAA
Pr001078668.1 / RSA000032339 / tgtaaaacgacggccagtTGGCGATTAAGATGGAAGGCA / caggaaacagctatgaccAATCCTTGCTGTTCCCACCCA
Pr001078670.1 / RSA000032343 / tgtaaaacgacggccagtCCAAGTCACGTAGCTTCTCTATTCGG / caggaaacagctatgaccCGGAAACCTGTGCGCCTG
Pr001078671.1 / RSA000032354 / tgtaaaacgacggccagtGGTCGAGGAAGTCACGTCGG / caggaaacagctatgaccACTTTGATCCATGGTCACAACTCA
Supplementary Table 2. Sanger sequencing of 16 CRC affected individuals identified 16 variants. One variant in the 5’ upstream region and another in the 3- UTR are novel. PTGS2 is transcribed from the antisense strand and the order of coordinates and reference nucleotide reflect this. The observed minor allele frequency (MAF) was calculated at each position by adding together the homozygous (Homo) and heterozygous (Het) allele dosage and dividing by the total number of DNAs with good quality sequence (Coverage).
Coordinate / rsID / Reference / Variant / MAF / Function / Coverage / Het / Homo / MAF
186,650,321 / rs20417 / G / S / 0.188 / 5' near gene / 16 / 3 / 0 / 0.094
186,650,211 / rs20418 / TAG / --- / 1.000 / 5' near gene / 16 / 0 / 16 / 1.000
186,649,757 / Novel / A / R / NA / 5' near gene / 16 / 1 / 0 / 0.031
186,649,618 / rs20424 / C / S / 0.012 / 5' near gene / 16 / 1 / 0 / 0.031
186,649,221 / rs2745557 / C / Y / 0.106 / intron / 15 / 3 / 0 / 0.100
186,649,004 / rs4648261 / G / R / 0.035 / intron / 15 / 1 / 0 / 0.033
186,648,197 / rs5277 / G / S / 0.165 / synonymous / 16 / 4 / 0 / 0.125
186,647,138 / rs4648268 / G / R / 0.071 / intron / 16 / 2 / 0 / 0.063
186,645,927 / rs2066826 / G / R / 0.159 / intron / 16 / 2 / 0 / 0.063
186,645,488 / rs4648276 / T / Y / 0.159 / intron / 16 / 2 / 0 / 0.063
186,644,282 / rs4648283 / G / R / 0.029 / intron / 15 / 1 / 0 / 0.033
186,643,058 / rs5275 / T / Y/C / 0.394 / 3' UTR / 13 / 8 / 2 / 0.462
186,642,950 / Novel / C / M / NA / 3' UTR / 14 / 1 / 0 / 0.036
186,642,429 / rs2206593 / C / Y/T / 0.053 / 3' UTR / 16 / 1 / 1 / 0.094
186,641,626 / rs2853805 / C / T / 1.000 / 3' UTR / 1 / 1 / 0 / 0.500
186,641,273 / rs689467 / T / K / 0.065 / 3' UTR / 16 / 3 / 0 / 0.094
Supplementary Table 3. Results of family-based association testing using the alternative trio-based algorithm.
Model / C allele / T allele / C allele / T allele / C allele / T allele
Additive / 64 / 64 / 0.6541 / 0.6541* / 0.5193 / 0.5193*
Dominant / 55 / 36 / 0.2678 / 0.4103 / 0.0588 / 0.0990
Recessive / 19 / 49 / 0.4103* / 0.2678* / 0.0990* / 0.0588*
Heterozygous-advantage / 64 / 64 / 0.1660 / 0.1660 / 0.0089 / 0.0089
*- A negative correlation was observed between the phenotype and the number of transmitted alleles
Supplementary Table 4. Results of family-based association testing using FBAT.
Model / C allele / T allele / C allele / T allele
Additive / 47 / 47 / 0.6946 / 0.6946
Dominant / 35 / 16 / 0.3180 / 0.4569
Recessive / 16 / 35 / 0.4569 / 0.3180
References
1. Yang H, Gu J, Lin X et al: Profiling of genetic variations in inflammation pathway genes in relation to bladder cancer predisposition. Clin Cancer Res 2008; 14: 2236-2244.
2. Cox DG, Buring J, Hankinson SE, Hunter DJ: A polymorphism in the 3' untranslated region of the gene encoding prostaglandin endoperoxide synthase 2 is not associated with an increase in breast cancer risk: a nested case-control study. Breast Cancer Res 2007; 9: R3.
3. Park JM, Choi JE, Chae MH et al: Relationship between cyclooxygenase 8473T>C polymorphism and the risk of lung cancer: a case-control study. BMC Cancer 2006; 6: 70.
4. Hu Z, Miao X, Ma H et al: A common polymorphism in the 3'UTR of cyclooxygenase 2/prostaglandin synthase 2 gene and risk of lung cancer in a Chinese population. Lung Cancer 2005; 48: 11-17.
5. Guo S, Li X, Gao M et al: Synergistic Association of PTGS2 and CYP2E1 Genetic Polymorphisms with Lung Cancer Risk in Northeastern Chinese. PLoS ONE 2012; 7: e39814.
6. Pu X, Lippman SM, Yang H, Lee JJ, Wu X: Cyclooxygenase-2 gene polymorphisms reduce the risk of oral premalignant lesions. Cancer 2009; 115: 1498-1506.
7. Lurie G, Terry KL, Wilkens LR et al: Pooled analysis of the association of PTGS2 rs5275 polymorphism and NSAID use with invasive ovarian carcinoma risk. Cancer Causes Control 2010; 21: 1731-1741.
8. Vogel U, Christensen J, Wallin H, Friis S, Nexo BA, Tjonneland A: Polymorphisms in COX-2, NSAID use and risk of basal cell carcinoma in a prospective study of Danes. Mutat Res 2007; 617: 138-146.
9. Li H, Ren C, Fan Z et al: A Genetic Variant in 3′-Untranslated Region of Cyclooxygenases-2 Gene Is Associated with Risk of Gastric Cancer in a Chinese Population. DNA Cell Biol 2012; 31: 1252-1257.
10. Danforth KN, Hayes RB, Rodriguez C et al: Polymorphic variants in PTGS2 and prostate cancer risk: results from two large nested case-control studies. Carcinogenesis 2008; 29: 568-572.
11. Mandal RK, Mittal RD: Polymorphisms in COX-2 Gene Influence Prostate Cancer Susceptibility in a Northern Indian Cohort. Arch Med Res 2011; 42: 620-626.
12. Zhu W, Wei BB, Shan X, Liu P: -765G>C and 8473T>C polymorphisms of COX-2 and cancer risk: a meta-analysis based on 33 case-control studies. Mol Biol Rep 2010; 37: 277-288.