Re: Role of the Oxidative DNA Damage Repair Gene OGG1 in Colorectal Tumorigenesis

Cheadle and colleagues (1) report that carrier status for a rare nonsynonymous variant in OGG1 (c.923G>A, rs113561019, G308E), confers susceptibility to colorectal cancer (CRC). Here, we report a well-powered study that finds no evidence for the association. This has important clinical genetics implications because inappropriate screening or intervention might be recommended to carriers, particularly those with a heterozygous MUTYH mutation, if OGG1 (c.923G>A) were wrongly ascribed as a CRC risk variant.

We analyzed 6856 case patients and 10090 control subjects from six European populations. These comprised 1) 3666 (predominantly) English case patients (n = 250 from the CORGI study; n = 957 from the QUASAR study; n = 1168 from NSCCG; n = 1291 from a Leeds-based case–control series) and 6140 control patients (n = 5694 1958 Birth Cohort [1958BC]; n = 446 from a Leeds-based case–control series); 2) 2052 Scottish CRC case patients and 2004 Scottish control subjects (n = 1452 from the 1935 and 1928 Lothian Birth Cohorts; n = 552 from Generation Scotland); 3) 276 Spanish case patients and 284 control subjects; 4) 800 Dutch samples (n = 337 Leiden case patients and n = 337 control subjects; n = 74 Gronigen case patients and n = 52 control subjects); 5) 199 Portuguese case patients and 186 control subjects; and 6) 1339 German samples (n = 77 Heidelberg case patients and n = 88 control subjects; n = 175 Kiel case patients and n = 999 control subjects). This provides greater than 99.9% power (α = 0.05) to detect the effect size (odds ratio [OR] =2.92) reported by Smith et al. (1).

We used Infinium HumanExome BeadChips (Illumina San Diego, CA) to genotype OGG1 c.923G>A (Supplementary Figure 1, available online) and validated genotyping by sequencing 541 randomly selected samples (n = 492 1958BC control subjects; n = 49 UK CRC case patients) showing complete concordance (r ² = 1.0; 531 both GG; 10 both GA). We used principal component analysis with HapMap 2 samples to assess ancestral comparability of case patients and control subjects (Supplementary Figure 2, available online).

None of the six series showed a statistically significant difference in frequency of c.923G>A genotype between case patients and control subjects (Figure 1; Table 1). In a meta-analysis of data from all studies, we found no association between c.923A carrier status and CRC (OR = 0.94; 95% confidence interval [CI] = 0.72 to 1.21; P = .61; P _heterogenity = .99; I ² = 0%) (Figure 1). Principal component analysis adjustment using Eigenstrat software had no impact on these findings (Supplementary Table 1, available online).

Figure 1.

Forest plot of association between OGG1 c.923A and colorectal cancer risk. Studies were weighted according to the inverse of the variance of the log of the odds ratio (OR) calculated by unconditional logistic regression. Meta-analysis under a fixed-effects model was conducted using standard methods. Cochran’s Q statistic to test for heterogeneity and the I ² statistic to quantify the proportion of the total variation due to heterogeneity were calculated. Horizontal lines indicate 95% confidence intervals (CIs). Boxes indicate odds ratio point estimate; its area is proportional to the weight of the study. Diamond (and broken line) indicates overall summary estimate, with confidence interval given by its width. Unbroken vertical line indicates null value (OR = 1.0). Risk allele frequencies are shown in square brackets.

Table 1.

OGG1 c.923G>A genotype counts and association statistics for the six colorectal case–control studies*

Study	Case					Control					OR (95% CI)	P
	AA	AG	GG	Total	MAF	AA	AG	GG	Total	MAF
England/Wales	1	39	3626	3666	0.0056	0	72	6068	6140	0.0059	0.93 (0.64 to 1.36)	.71
Scotland	0	21	2031	2052	0.0051	0	23	1981	2004	0.0057	0.90 (0.52 to 1.56)	.72
Holland	0	6	405	411	0.0073	0	5	384	389	0.0064	1.07 (0.43 to 2.64)	.89
Spain	0	4	272	276	0.0072	0	5	279	284	0.0088	0.91 (0.35 to 2.38)	.84
Portugal	0	4	195	199	0.0101	0	6	180	186	0.0161	0.75 (0.29 to 1.95)	.56
Germany	0	3	249	252	0.0060	0	16	1071	1087	0.0074	1.17 (0.47 to 2.92)	.74
Combined	1	77	6778	6856	0.0058	0	127	9963	10090	0.0063	0.94 (0.72 to 1.21)	.61

* Odds ratios and P values were generated using logistic regression. For the combined analysis, odds ratio and P value shown are from fixed-effects meta-analysis.

There are a number of possible explanations for the disparity of our findings and those reported by Smith et al. (1). First, population stratification could lead to spurious associations, especially with rare variants. Indeed, we observed a higher frequency for c.923G>A in Spanish and Portugese populations than in other populations (Table 1). Smith et al. (1) largely relied on self-reported ancestry data, whereas we ensured ancestral comparability of case patients and control subjects from single nucleotide polymorphism genotypes, thereby excluding this as a source of bias. Second, small sample size may generate spurious false-positive associations. The reported odds ratio for the association between c.923G>A was 2.92, but we had greater than 90% power to detect an association with odds ratio of 1.80 [lower 95% CI reported by Smith et al. (1)]. One biological explanation is that case patients analyzed by Smith et al. (1) were selected for advanced disease. Hence it is formally possible that a relationship only exists between genotype and aggressive CRC.

In conclusion, in this well-powered study, we find no evidence to support OGG1 c.923A as a CRC susceptibly allele.

Funding

The Leeds collection was supported by Cancer Research UK Programme Award (C588/A10589). MGD’s laboratory is supported by a Cancer Research UK programme grant (C348/A12076). Funding for quality control of the 1958 Birth Cohort exome chip data was through the Wellcome Trust Case Control Consortium. The Portuguese study was partly funded by the Genetic study of Common Hereditary Bowel Cancers in Hispania and the Americas (CHIBCHA) FP7 project and by Liga Portuguesa Contra o Cancro, Núcleo Regional do Norte. Work in IT’s laboratory is supported by funding from the Oxford National Institute for Health Research Biomedical Research Centre and core funding to the Wellcome Trust Centre for Human Genetics from the Wellcome Trust (090532/Z/09/Z). Work in the laboratory of RSH is supported by funding from Cancer Research UK (C1298/A8362, supported by the Bobby Moore Fund). Additional funding is provided by the National Health Service through the Biological Research Centre of the National Institute for Health Research at the Royal Marsden Hospital National Health Service Trust. BK is in receipt of a PhD Studentship from the Institute of Cancer Research and receives sponsorship from the Sir John Fisher Foundation. SCB is supported by a contract from the Fondo de Investigación Sanitaria (CP 03-0070). Centro de Investigación Biomédica en Red en el Área temática de Enfermedades Hepáticas y Digestivas (CIBERehd) and Centro de Investigación Biomédica en Red de Enfermedades Raras (CIBERER) are funded by the Instituto de Salud Carlos III. This work was supported by grants from the Fondo de Investigación Sanitaria/Spanish Federation of Rare Diseases (11/00219, 11/00681), Asociación Española contra el Cáncer (Fundación Científica GCB13131592CAST), and Fundació Olga Torres (SCB and CRP). We are sincerely grateful to all patients participating in this study who were recruited in 25 (EPICOLON 1) and 14 (EPICOLON 2) Spanish hospitals as part of the EPICOLON project. We are also indebted to the University of Santiago de Compostela and University of Pompeu Fabra nodes of Spanish National Genotyping Center (El Centro Nacional de Genotipado y el Instituto de Salud de Salud Carlos III), the Genomics Unit of Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS) and the Biobanks of Hospital Clínic – IDIBAPS for technical help. The work was carried out (in part) at the Esther Koplowitz Centre, Barcelona.

We acknowledge use of control genotypes from samples from the Lothian Birth Cohort and Generation Scotland. The 1958 Birth Cohort exome chip data was QCd by Kathy Stirrups. Data sharing was organized by the UK Exome-chip consortium. Exome-chip design information: http://genome.sph.umich.edu/wiki/Exome_Chip_Design. Exome variant server: http://evs.gs.washington.edu/EVS/.