Chronic inflammation has demonstrated an important role in cancer (12
); tumor microenvironments of numerous cancers, including ovarian, have shown high-levels of inflammatory proteins, such as cytokines, and aberrations in interleukin 1 alpha, interleukin 6, and tumor necrosis factor alpha (13
). Thus, examinations of genetic variants in these pathways may shed light on inflammation-related carcinogenesis. In the largest study to date, we examined the relationship between SNPs in inflammation-related genes and risk of ovarian cancer using a two-study discovery analysis followed by examination of five studies participating in the Ovarian Cancer Association Consortium. We found that minor alleles at rs17561 and rs4848300 in IL1A
and rs1864414 in ALOX5
were consistently inversely associated with risk of ovarian cancer, and, intriguingly, that the IL1A
associations were limited to rarer histological subtypes.
Interleukin 1 alpha is a cytokine which is involved in numerous immune responses and inflammatory processes; it is produced by monocytes and macrophages and released in response to cell injury for the induction of apoptosis. A recent analysis of 4,552 SNPs in cytokines reported that IL1A
rs17561 is one of only eight SNPs leading to a predicted intolerable amino acid change, that it is by far the most widely studied cytokine SNP, and that minor alleles here have been “strongly” associated with risk of numerous diseases involving an inflammatory response such as allergic rhinitis, atopy, nasal polyposis, malaria, gingival hyper-inflammatory response, systemic sclerosis, severity of periodontitis, antibody responses to periodontal microbiota (14
) and, more recently, ankylosing spondylitis (15
). While we are the first to report an association with ovarian cancer risk, a meta-analysis of prospective breast cancer studies has reported increased risk with IL1A
rs17561 (p=0.008) as well as modification of the dose-response relationship between cumulative personal diagnostic radiation and breast cancer risk (interaction p=0.004) (16
). Another study of breast cancer risk found that IL1A
was the most significant of 232 inflammation genes in gene-level and haplotype analyses (p=4 × 10−4
). Of note, a prior ovarian cancer study of approximately 300 participants in total found no association between risk and other SNPs in IL1A
; they did not examine rs17561 or rs4848300 and were underpowered to detect associations of the type we observed (18
Epithelial ovarian cancer has several histologic subtypes with characteristic molecular signatures, cytogenetic features, oncologic signaling pathways, and clinical behavior. Nonetheless, for maximal sample size and smaller numbers of tests, molecular epidemiologists have traditionally combined cases and performed subtype analyses only on SNPs associated with risk overall, as we did here. However, our IL1A findings demonstrate that, if multiple rarer subtypes share a common genetic etiology, associations with rarer subtypes (rather than the most common and most studied serous subtype) may drive the result among all cases. These results underscore the need for additional candidate gene or genome-wide studies of clear cell, mucinous, and endometrioid ovarian cancer.
Arachidonate 5-lipoxygenase is a member of the lipoxygenase gene family assisting with conversion of arachidonic acid to leukotrienes, which are important mediators of a number of inflammatory and allergic conditions; it is expressed in the ovarian cancer (8
). Interestingly, we previously reported an association with poorer survival and ALOX5
rs2115819 and rs12264801 among the MAY cases studied here (serous HR 1.54, 95% CI 1.18-2.02, p=0.002; HR 1.31, 95% CI 1.02-1.69, p=0.04, respectively) (19
). Even though rs1864414 was not associated with ovarian cancer survival (p=0.23) (19
), we evaluated whether the minor allele may be under-represented among cases studied here leading to a false inverse association with risk. We saw no association with rs1864414 genotype and time to enrollment and no difference in risk estimates by mean time of enrollment, and therefore don’t expect that the inverse association reported here is due to survival bias.
In addition to a large sample size (7,212 participants), strengths of this report include inclusion of numerous genes (most comprehensive to date), use of the tagSNP approach, and independent analyses of discovery and replication datasets. The latter issue requires some consideration. Because false discovery is a concern in SNP association studies, we sought replication of initial results in a large independent dataset. Some argue for the routine analysis of the likelihood of false positive results. With one commonly used method (20
), the IL1A
associations are “noteworthy” (probability of a false positive result < 0.2) only if the prior probability of an association was 0.1 or greater. Of course, choice of the appropriate value for prior probability is a matter of great debate. Some argue that in a discovery-replication study, prior probabilities of association is high, given that significant results were seen in discovery analysis; others argue that in a candidate gene study, biological plausibility is increased. Regardless, additional replication is the key to revealing the “truth” of these associations.
Limitations of our analyses include lack of genotypic coverage of uncommon variants, assessment of epistasis or gene-environment interactions, and multi-SNP IL1A analysis. Additional genotyping or next generation sequencing efforts are needed to target additional variants, and, due to the need for greater statistical power, larger samples sizes are needed to address uncommon variants as well as interactions. A larger study would also enable simultaneous examination of multiple SNPs in IL1A and the related chromosome 2 cytokine gene cluster. We also note that the boundaries of biological pathways, such as inflammation, continue to evolve; thus expansion of genes from this set of 27 genes will be informative in future efforts. Finally, incorporation of genotype-phenotype studies considering quantification of tumor immune cells, tumor mRNA or protein expression, or inflammatory serum biomarkers into studies of inflammation SNPs will enhance these projects as well.
In conclusion, SNPs in ALOX5 and IL1A appear to harbor common inherited variants associated with modest differences in risk of ovarian cancer. Efforts to expand current data on inflammation and this deadly disease are a high priority.