Presenter: Emilio Alonso MD | Providence Portland Medical Center Background: Pancreatic ductal adenocarcinoma (PDAC) continues to be one of the leading causes of cancer death in the Unites States, with an increasing incidence over the past decades. Despite the high mortality rate and dismal prognosis, the therapeutic responses are poor and heterogenous even to our current standards of treatment. Recently it has been shown that PDAC could be subdivided into clinically relevant subgroups using immunohistochemistry (IHC) for two protein markers, KRT81 and HNF1A. We aimed to validate their IHC-subtyping and evaluate the gene expression HNF1A and KRT81 in an upfront-resected PDAC cohort. Alternative PDAC subtyping using gene-expression profiles have been previously described by Collisson and Bailey, which we aim to see their overlap using our cohort. Methods: We performed IHC for HNF1A and KRT81 on 61 patients with upfront resected PDAC from 2010 to 2014. These were read by a board-certify pathologist confirming its presence or absence in the tumor cells. Of the 68 patients, 34 patients had RNAseq. Gene expression analysis for KRT81/HNF1A were generated, from which gene expression ratio were calculated. Expression of HNF1A/KRT81 in the The-Cancer-Genome-Atlas (TCGA) pancreatic cancer database, were similarly evaluated on 138 patients after the data was obtained from cBioPortal. Using RNAseq as a more sensitive test, the accuracy, sensitivity, and specificity of IHC-subtyping was accessed. Molecular subtyping was also performed as previously described by Collisson and Bailey. Results: Using IHC-subtyping, we were able to validate the Results of the previous study. The IHC HNF1A-subtype showed a better overall and disease-free survival than KRT81. Similarly, the HNF1A positive RNAseq subtype had a statistically better overall survival and disease-free survival over the KRT81 positive subtype (P < .016 and P < 0.023, respectively). When validated externally using the TCGA-database, the RNAseq HNF1A subtype had a statistically better disease-free survival than the KRT81 subtype (P < 0.0193). We were not able to recapitulate this on the overall survival given the early censorship of the TCGA cohort of patients. Using gene expression as a more sensitive tool, the accuracy (53%), sensitivity (54%) and specificity (75%) were calculated. Thus, IHC had a better positive than negative predictive value (84% vs 6%) in stratifying PDAC. Additionally, our gene profile subtypes did not match the previously described subtypes of Collisson and Bailey as they did not correlate with survival outcomes in our patients. Conclusion: RNAseq stratification was able to accurately recapitulate the IHC subtypes of PDAC using the gene expression of HNF1A and KRT81, which was associated with significant different outcomes in both our internal cohort of upfront resected PDAC and externally using the TCGA database. Unfortunately, we were not able to recapitulate the previous molecular subtyping defined by Collisson and Bailey. This study, introduces a more sensitive analytical tool, RNAseq, for the stratification of this PDAC subtypes, which may play a role in our future understanding of the biology of these two subgroups. With the growing development in price and workflow, RNAseq analysis on endosonographically acquired core needle biopsies may serve as a biomarker-based stratification stool for PDAC patients.
Read full abstract