In an absolutely remarkable, fascinating and possible game-changing study, Joo Mi Yi and Nita Ahuja and colleagues (primarily) from Johns Hopkins University have identified a “panel” of biomarkers based on DNA methylation of two relatively obscure genes (that can be identified in serum) for use, at least at an experimental level at this stage, for the earlier detection of early pancreatic cancer.
In this area of earlier detection research, there has been large attention given to discovering genetic DNA mutations that are common to pancreatic cancer tumors. For example, about a year ago in the journal Nature (November 15, 2012) an international group of researchers aiming for a comprehensive listing of such mutations identified 16 significantly mutated genes with 2,016 mutations and many other genetic variations. We commented on this study here on Pancreatica in a 12/09/12 blog entry.
Another area of recent active such research, for example, has been in the arena of MicroRNAs. And notably at Johns Hopkins (along with other institutions) there has been a consistent but less heralded look at DNA methylation changes that occur in pancreatic cancer. Methylation is the natural process (as related to DNA in mammals) whereby a methyl group is found at cytosine-phosphate-guanine (CpG) sites on DNA, effectively tending to silence the activity of the underlying referent gene. About 60% to 90% of CpG sites are methylated in mammals. So, one key point in looking for such changes is to seek sites which ARE methylated in tumor conditions, but which are NOT methylated with normal tissue.
Yi, Ahuja and colleagues electronically published the results of their elaborate research on November 1, 2013 in Clinical Cancer Research, the official journal of the American Association for Cancer Research. Using cell lines as well as human samples, they narrowed methylated gene candidates down from 1,427 genes to eight that showed methylation in pancreatic cancer. Of these eight, they identified two that were the two most methylated genes: BNC1 (91% frequency) and ADAMTS1 (67% frequency). These genes have heretofore not been particularly associated with pancreatic cancer. And it is not entirely clear what their full functions are – although BNC1 appears to have a tumor suppressor role, and ADAMTS1 may be involved in angiogenesis. Both of these genes showed “dense” methylation in cell lines and in pancreatic cancer, and showed almost no methylation in normal pancreatic samples. Also, these genes did not tend to demonstrate increased methylation in pancreatitis.
Additionally (importantly), the authors used a fairly newly developed very sensitive “nano-enabled” assay to test the serum of patients diagnosed in various stages of pancreatic cancer (including the possible precursor to pancreatic adenocarcinoma: pancreatic intraepithelial neoplasia), and as compared to CA 19-9 levels. The rates of methylation increased at every stage of disease, and demonstrated higher rates than those of CA 19-9 (until stages III and IV where methylation and CA 19-9 levels were both 100%).
The overall sensitivity of this two-gene methylation “panel” was a respectable 81%; the specificity was 85%. Thus, this panel appears to represent a highly promising approach aimed at the earlier diagnosis of early pancreatic cancer. Further research is required to verify and expand the findings, but these results are indeed encouraging.
Dale O’Brien, MD