Genome-Wide Association Study (GWAS) of Renal Cell Carcinoma (RCC)

IARC: Ghislaine Scélo, James McKay, Mattias Johansson, Paolo Boffetta, Paul Brennan
NCI: Stephen Chanock, Lee Moore, Mark Purdue, Nat Rothman
CNG: Marc Lathrop

N/A

The vast majority of cancers that arises in the kidney are renal cell carcinomas (RCC) of the parenchyma: they can be classified as clear cell carcinomas (approximately 80% of all kidney cancers), papillary carcinomas, chromophobe carcinoma, carcinomas of the collecting ducts of Bellini, and medullary carcinomas (Scélo and Brennan, 2007, Eble JN et al. (Eds; 2004)). Other types, such as sarcomas or transitional cell carcinomas of the renal pelvis, are comparatively rare in adults (approximately 15%) and will not be integrated in our project.

A sharp increase in the incidence of RCC was observed in numerous registries reported in the two volumes of the Cancer Incidence in Five Continents series for the periods 1983-87 and 1993-97 (Parkin DM et al. (Eds; 1992), Parkin DM et al. (Eds; 2002)). Some of the greatest increases were observed in the Czech Republic and among the black population in the US. These trends towards increased incidence are unlikely to be fully explained by increased detection of presymptomatic tumors; instead, they probably reflect – to an unknown degree -real increases in the numbers of new cases (Hock LM et al. (2002)).

Worldwide, the male:female ratio in incidence of RCC is 1.6:1.0 and in both men and women, annual incidence rates of RCC show a linear and regular increase up until 75 years of age, when they reach a plateau (Ferlay et al. (Eds; 2004), Parkin et al. (Eds; 2005)). Geographical variations in annual incidence rates (age-standardized for the world population) are shown in Figure 1. In both sexes, the highest (country-wide) annual incidence rate occurs in Czech Republic (21.1 per 100,000 in men and 10.2 per 100,000 in women) and the lowest rates are found in Africa and Asia. In the US, RCC is more frequently diagnosed in the black population (age-standardized annual incidence rates of 13.3 per 100,000 in men and 7.1 per 100,000 in women) than in the white individuals (11.1 per 100,000 in men and 5.6 per 100,000 in women).

There is some evidence of an ongoing improvement in survival after kidney cancer diagnosis in the US. For example, Chow et al. (1999) reported an increase in 5-year survival in US patients diagnosed as having any stage of kidney cancer, but only in the white population. However, survival after kidney cancer diagnosis remains low, ranging from 78% when tumor was diagnosed at stage 1 to as low as 7% when diagnosed at stage 4 (Figure 2). This low survival rate is striking when considering that in 2003 in the US, almost one case of kidney cancer out of five was diagnosed at stage 4 (National Cancer Database (online 30 October 2006)[http://www.facs.org/cancer/ncdb/] (accessed 21 February 2007)).

Based on current knowledge, the main avoidable causes of kidney cancer are cigarette smoking, excess body weight and hypertension, which together are likely to account for up to 60% of all cases of these tumors on average (Scélo and Brennan, 2007). However, mechanisms involved are yet to be elucidated. A substantial proportion of kidney cancers are also likely to be related to diabetes, although further information on whether diabetes is an independent risk factor for kidney cancer is required. Growing evidence for a role of vegetable intake is intriguing and deserves further consideration (Lee et al., 2009, Moore et al., 2008, Hsu et al., 2007). Implication of folate is a promising hypothesis, considering that fresh vegetables are the main source of folate (Allen, 2008), some genes involved in the one-carbon metabolism have
been associated with RCC risk (Moore et al., 2008), and low folate is thought to increase the risk of colorectal cancer (Ulrich and Potter, 2007). Most previous studies on cancer and folate relied on diet questionnaires to estimate the intake, a method which suffers from substantial misclassification. Instead, we propose to measure plasma concentration of folate and other metabolites of the pathway to (i) reduce measurement errors, (ii) take into account other causes of deficiency, such as malabsorption, and (iii) have a comprehensive assessment of metabolites involved in the pathway (Figure 3).

We expect our results to provide new candidate risk factors (genetic) and confirm suspected risk factors (low folate intake), and to help understanding the mechanisms involved in renal cell carcinogenesis and kidney cancer survival.

N/A

Renal cell carcinoma (RCC) is the 8th most common cancer in the US and the 10th most common form of cancer death, with a particularly high incidence among African Americans. Apart from smoking, obesity and hypertension, much of the etiology of this disease remains to be elucidated. A joint US/European genome-wide association (GWA) study is currently in progress to identify new loci associated with RCC risk, involving approximately 4,000 RCC cases (1,400 from cohort studies) and 9,000 controls (6,000 from cohort studies). It is organized by IARC and the US NCI, and includes the collaboration of five cohorts (three in Europe and two in US) and two large case-control studies (one in Europe and one in US). An additional case-control study is currently underway in Europe and 1,500 case-control pairs will be added to the above numbers in the next future. Provisions are being made to organize a consortium of case-control studies of RCC which will provide the framework for future replication studies.

To accelerate the understanding of the mechanisms underlying RCC, we propose the following specific aims, which would complement current case control-based work:
1. To recruit at least 1,000 cases and 2,000 controls from additional or updated cohorts;

2. To conduct whole-genome genotyping (at least 610k SNPs) on these additional cases and controls in order to expand our existing GWA analyses to over 6,000 cases and 12,000 controls;

3. To obtain survival information on at least 4,000 from the 6,000 cases recruited in case-control and cohort studies, in order to investigate the association between RCC survival and genetic variants, and identify whether variants associated with the disease onset are also associated with survival;

4. To measure the plasma concentrations of B vitamins and other metabolites involved in the one-carbon metabolism in prospectively collected blood samples from at least 1,000 cohort-based cases and 2,000 nested controls, and to analyze their association with RCC risk (overall and after consideration of variants in the genes involved in the one-carbon metabolism pathway).

1. Participants: At least 1,000 pathologically confirmed incident cases of RCC (ICDO-3 C64) diagnosed among non-Hispanic Caucasians (and twice as many controls of the same ethnicity) with sufficient DNA for scanning will be included in the project from among the participating cohort studies. Following a nested case-control design, controls will be randomly selected and individually matched on cases for sex, age (± 2 years) and date of blood draw (± 3 months) within each participating cohort using the incidence density sampling method. The sample will be supplemented by at least 1,000 cases and controls from an ongoing population-based case control
study in Central Europe for the GWAS part of the project.

2. Genotyping laboratory methods and quality controls: We will genotype >610,000 SNPs across the genome using the Illumina 610-Quad array in two different labs: the Centre National de Génotypage (CNG, Paris, France) and the Core Genotyping Facility (CGF, NCI). IDAT files from each study will be requested to enable SNP reclustering. The quality of DNA samples will be assessed on the basis of sample call rates. The quality of the genotype data will be assessed on the basis of the call frequency rate, concordance rate between blind duplicate DNAs, the frequency of Mendelian errors in CEPH families (added as controls) and the distribution of the significance of the Hardy and Weinberg disequilibrium constant in the control group. In addition we will check for chromosome X heterozygosity among men, unexpected duplicates and relatedness. SNPs classified unreliable by these indicators will be removed from the association analysis. Quality control assessment will be conducted at the NCI Core Genotyping Facility, CNG and IARC.

3. Biochemical analyses: The study will include measurements of plasma concentration of vitamins B9 (folate), B6 (pyridoxal phosphate), B2 (riboflavin), and B12 (cobalamin), as well as choline, betaine, methionine, cysteine, and homocysteine (Figure 3). For that purpose, 0.5ml of plasma will be requested for each participant. Measurements will be conducted on serum samples if plasma is not available. All assays will be done by laboratory personnel blinded to the case-control status of the blood samples. Various methods of measurement will be used . Within- and between-run variations will be calculated to assess the accuracy and
repeatability of the measurements.

4. Statistical analyses: For the GWA part of the study, association tests with RCC risk (Aim 2) will be conducted using unconditional logistic regression modeling assuming log-additive SNP effects. The updated dataset will comprise at least 6,000 cases and 12,000 controls. All models will include covariates for age, sex, participating study, and principal components capturing population substructure; other covariates to be potentially adjusted for in some models include body mass index, blood pressure, smoking status (current, former, never), smoking frequency and family history of cancer. Our criterion for statistical significance is an alpha of 1.0x10-7. SNPs that surpass or closely approach this alpha will be regenotyped in all samples by Taqman assays.

Association with survival (Aim 3) will be conducted in the same way, except that Cox modeling will be used instead of logistic regression. The following outcome will be included in the analyses: relapse, second primary cancer, death.

Plasma concentration of B-vitamins and other analytes will be categorized into quartiles with cut points based on the concentration distribution in the controls subjects. Odds ratios for plasma concentrations will be calculated with unconditional logistic regression techniques, using the same standard covariates as above. Stratified analyses by delay between blood draw and case diagnosis will also be conducted.

STATISTICAL POWER
With the planned sample size (6,000 cases and 12,000 controls) and a p-value threshold of 1.0x10-7 for the GWAS, a statistical power of 80% will be reached to detect odds ratio of 1.35 and 1.15 for minor allele frequencies of 0.05 and 0.40, respectively.

For biochemical analyses (1,000 cases and 2,000 controls), a statistical power of 80% will be reached to detect odds ratio of 1.37 between two quartiles with a p-value threshold of 0.05.

N/A

N/A

N/A

N/A

N/A

Yes