Lack of Association for Reported Endocrine Pancreatic Cancer Risk Loci in the PANDoRA Consortium

Pancreatic neuroendocrine tumors (PNETs) are rare neoplasms, but their incidence has greatly increased in the past decades (1). In comparison with other solid tumors, little is known about PNET risk factors, and only a handful of association studies have been performed to uncover the genetic determinants of the disease (2–4). Berkovic and colleagues have performed studies on inflammation-related genes, such as ILB1 and TNFA (2, 3), while Ter-Minassian and colleagues have selected a more comprehensive approach using a custom array containing almost 1,500 SNPs (4). Both teams suggested several interesting associations; however, due to the capricious nature of association studies and the relatively small sample size of these PNET studies, it is of uttermost importance to validate their findings in an independent population, such as the PANcreatic Disease ReseArch (PANDoRA) study (5).

The PANDoRA consortium has been described elsewhere (5). In this study, 344 PNET cases and 2,721 controls have been genotyped (Table 1). Cases were sporadic, that is, not observed in the context of genetic syndromes associated with PNET, such as multiple endocrine neoplasia (MEN)-1, MEN-2, Von Hippel–Lindau or tuberous sclerosis. Controls were selected in the same geographic areas as the cases. We selected 16 SNPs that represent all the polymorphic variants so far identified as risk loci for PNETs but not replicated yet. In addition to the SNPs reported in this article, we have recently performed a study on the CDKN2A gene variability in relation to PNET susceptibility, and therefore, all the variants of this gene are not reported here (6). Genotyping was performed using the KASPar SNP genotyping system (KBiosciences). The order of DNA samples from cases and controls was randomized on plates to ensure that similar numbers of cases and controls were analyzed simultaneously. For quality control purpose, around 8% of the samples were duplicated and genotype concordance was checked. Statistical analysis was performed using an unconditional logistic regression setting the more common allele for each polymorphism as reference and adjusting for age, gender, and country of origin. Given that in the original article on IL1B (3), an analysis was performed combining the alleles of the two SNPs, we also performed it (using the same criteria as Berkovic). We also used online bioinformatic tools such as HaploReg (http://www.broadinstitute.org/mammals/haploreg/haploreg.php), RegulomeDB (http://www.regulomedb.org/), and GTEx (http://www.gtexportal.org/home/) to explore potential functions of the SNPs.

The average call rate of the 16 genotyped SNPs was 96.63% (92.06–99.04), and the concordance rate of duplicated genotypes was higher than 99%. The genotype distributions at all loci were in Hardy–Weinberg equilibrium in controls, with nonsignificant χ² values. Table 2 shows the frequencies and distribution of the genotypes, the ORs and 95% confidence intervals for the association with PNET risk. None of the SNPs showed any statistically significant association considering a P value threshold of 0.05. The only potentially interesting effect was a trend for the carriers of the minor allele (A) of the IL1B-rs1143634 SNP and a decreased risk of developing PNET (P_homozygous = 0.082). The combined diplotype analysis for the IL1B gene, as done in the original article by Berkovic and colleagues (3), did not reveal any statistically significant associations (data not shown). For IL1B-rs1143634, HaploReg showed two possible eQTLs (with expression of gene CHCHD5, P = 4.1 × 10⁻⁵, and of gene SLC20A1, P = 0.0014), RegulomeDB showed a score of 5 (minimal binding evidence), and GTEx showed no statistically significant results.

Because of the rarity of PNETs, there have been only a small number of studies investigating the genetic susceptibility to this disease. The most interesting findings have been polymorphic variants in genes involved in inflammatory response, cell-cycle control, and DNA repair mechanisms. This is one of the largest studies on PNETs to date, and our sample size was by far larger than those used in the original studies, giving us more than 95% statistical power to detect the previously reported associations. The IL1B-rs1143634 variant is a synonymous, possibly functional, SNP that has been widely studied in relation to a variety of human diseases and conditions, making it an attractive candidate for PNET risk. Our results however suggest, at best, only a minor influence of the variant in the etiology of the disease. In addition, we found the minor allele to be associated with decreased risk, whereas in the original publication, it was the opposite. The three selected bioinformatic tools did not reveal a clear-cut indication about the functional effect of the SNP. In conclusion, in this study, we can exclude a major role of the selected polymorphisms in PNET etiology and highlight, on one hand, the importance of finding genetic markers for the disease, ideally through a genome-wide association study approach, and on the other hand the need of replication of epidemiologic findings in independent populations.

No potential conflicts of interest were disclosed.

Conception and design: D. Campa, E. Costello, O. Strobel, J.P. Neoptolemos, G. Capurso, F. Canzian

Development of methodology: M. Pastore, J.P. Neoptolemos

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): O. Obazee, F. Panzuto, V. Liço, W. Greenhalf, R. Talar-Wojnarowska, O. Strobel, C.F. Zambon, J.P. Neoptolemos, G. Zerboni, R. Kaaks, T.J. Key, C. Lombardo, K. Jamroziak, D. Gioffreda, T. Hackert, K.-T. Khaw, S. Landi, A.C. Milanetto, L. Landoni, R.T. Lawlor, F. Bambi, F. Pirozzi, D. Basso, C. Pasquali, G. Capurso

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Pastore, F. Canzian

Writing, review, and/or revision of the manuscript: D. Campa, O. Obazee, F. Panzuto, V. Katzke, E. Costello, R. Talar-Wojnarowska, O. Strobel, J.P. Neoptolemos, R. Kaaks, T.J. Key, K. Jamroziak, T. Hackert, K.-T. Khaw, G. Capurso, F. Canzian

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): O. Obazee, M. Pastore, V. Liço, W. Greenhalf, F. Tavano, V. Corbo, O. Strobel, C. Lombardo, T. Hackert, K.-T. Khaw

Study supervision: T. Hackert, F. Canzian

This work was partially supported by CRUKC8221/A19170 and MRCMR/M012190/1 grants.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.