When we look at pancreatic cancer, even after years of research development, we still see a bleak picture. Its incidence is increasing in many countries, including Japan, for reasons as yet unexplained. Early detection is difficult and the five-year survival proportions remain the worst (<10%) among major malignancies. The staggering fact is that while the prognosis for most cancers is counted in years, for pancreatic cancer it is counted in mere months.
My first encounter with pancreatic cancer occurred more than 20 years ago, during my PhD at the Department of Preventive Medicine, Nagoya University. In a population-based case-control study of pancreatic cancer, we enrolled 242 patients. One patient among them has lingered in my mind. I first saw her during inpatient rounds. Admitted with stage 4 pancreatic cancer, she died within 3 months of the diagnosis. She was only 27 years old at that time. It was this encounter that played a big part in my decision to choose pancreatic cancer etiology research as the core of my career. I still vividly remember her haggard face in her final days, and my own slow, sinking feeling of helplessness in those moments. Those haunting memories are what motivate me to persevere in my pancreatic cancer research.
The idea of conducting a genome-wide association study (GWAS) of pancreatic cancer dates back to 2011. Around that time, I was stuck in a rut, researching a long list of lifestyle factors that might contribute to pancreatic cancer. Fascinated by the power of GWAS in discovering disease-causing genes, I decided to change gears and focus on the role of inherited genetic variations in cancer susceptibility. What is amazing about GWAS is that it is hypothesis-free, meaning that an unexpected, lucky find is possible. Such a serendipitous discovery can sometimes point directly to a radical, new insight into disease mechanisms. On the other hand, GWAS faces its own limitations. About 90% of GWAS hits are located in either intron or intergenic regions of the genome; the underlying mechanisms are not clear. And the vast majority of SNPs exert modest effects, with an OR of 1.1-1.3. At that time in Japan, aggregation of such a large number of cases and controls for a GWAS of pancreatic cancer, followed by functional validation of GWAS hits, has never been done before.
To meet this challenge, we formed a multidisciplinary research group (now JaPAN consortium). Mainly due to the efforts of physician scientists, we were able to collect a large number of samples in a relatively short period. By combining the data with those from a Hospital-based Epidemiologic Research Program at Aichi Cancer Center (HERPACC; PI: Keitaro Matsuo), we included 943 cases and 3597 controls in the initial JaPAN GWAS. Perhaps unsurprisingly, our first attempt did not yield any genome-wide significant SNP associations. However, we remained undaunted and immediately reached out to two other research groups (BBJ and NCC) asking for collaborations on a meta-analysis of GWAS data. As a result, we finally assembled 2,039 cases and 32,592 controls, the largest sample size in the Asian population. In addition, replication went well too; the top SNP association was replicated in six additional studies from both Japan and abroad.
Among many interesting findings from this study, I would like to highlight three. First, the lead variant rs78193826 maps to the exon of the protein-coding GP2 gene. The OR was approximately 1.5 for individuals with the T risk allele compared with those with the alternative C allele. The association of this SNP with pancreatic cancer may be the strongest among identified SNPs so far.
Secondly, it turns out that rs78193826 is an Asian-specific variant; the minor allele frequency (MAF) of T is 3-8% in Asian populations, while it is <0.01% in populations of European ancestry.
Third, the encoded GP2 protein is abundantly expressed in the pancreas. This suggests that genetic variations in the GP2 gene locus may be associated with pancreatic diseases.
Our success in discovering the GP2 variants associated with pancreatic cancer in the Japanese population demonstrates the importance of expanding GWAS to understudied, non-European populations. This feat is wholly attributed to the power of our multidisciplinary team science; Masahiro Nakatochi of Nagoya University did painstaking SNP imputations and GWAS meta-analyses; Yasuyuki Hosono and Issei Imoto of Aichi Cancer Center performed state-of-the-art functional characterizations; and so many co-authors from different fields contributed to each part of this project.
Big challenges still lie ahead. Although several lines of evidence suggest that the missense variant rs78193826 is the "smoking gun”, it does not necessarily mean that it is the only causal variant of the GP2 gene. We need to do fine-mapping and follow-up wet experiments in order to establish the causal relationship. The ongoing functional characterization for this variant has already yielded some tantalizing evidence, as presented in the paper. What I found most interesting is that the functional relevance may be related to the modulation of KRAS activity, the most frequently mutated driver gene in early pancreatic carcinogenesis.
These fortuitous findings on GP2 variants open up several avenues for further exploration of pancreatic cancer etiology. It is truly rewarding to navigate the path from association derived from thousands of patients, to elucidation of underlying mechanisms, for the benefit of each and every person suffering from pancreatic cancer.
- Wojcik GL, et al. Genetic analyses of diverse populations improves discovery for complex traits. Nature 2019;570:514-518.
- Hase K, et al. Uptake through glycoprotein 2 of FimH+ bacteria by M cells initiates mucosal immune response. Nature 2009;462:226–230.
- Cogger KF, et al. Glycoprotein 2 is a specific cell surface marker of human pancreatic progenitors. Nat Commun 2017;8:331
- Suzuki K, et al. Identification of 28 new susceptibility loci for type 2 diabetes in the Japanese population.Nat Genet 2019;51:379-386.