M. C. Hernandez1, J. Leiting1, L. Yang2, J. R. Bergquist1, M. J. Truty1 1Mayo Clinic,Department Of Surgery,Rochester, MN, USA 2Center For Individualized Medicine,Biomarker Discovery Program,Rochester, MN, USA
Introduction:
Adenosquamous carcinoma of the pancreas (ASCP) is a rare and lethal histologic subtype of pancreatic cancer. ASCPs are defined by a mix of at least 30% malignant squamous cell carcinoma and coexisting ductal adenocarcinoma. Normal pancreas tissue has no benign squamous epithelial components. Thus the origin of this tumor is uncertain. Postoperative recurrence rates are high and ASCP demonstrates significantly worse overall survival, even compared to ductal adenocarcinoma. The low prevalence of ASCPs makes research studies and clinical trials exploiting unique features of this tumor difficult. We aimed to generate and amplify patient-derived ASCP malignant tissue in order to (1) genomically characterize (2) functionally assess for sensitivity to MTH1 inhibition and (3) correlate functional assay sensitivity with therapeutic response using tumor bearing ASCP mice.
Methods:
Patient derived xenografts (PDX) were generated from surgical resection of patient tumor tissue in NOD SCID mice. All patient and derived PDX tumors were histologically (H&E and IHC) confirmed. We performed whole genome mate pair sequencing (MPseq) on PDX tissues. Western blot and immunohistochemistry for the presence of MTH1 enzyme was performed to identify possible sensitivity to MTH1 inhibition. Cells were cultured using a hanging drop technique and treated with cytotoxic and targeted therapies. Cell viability was assessed using daily cell counts and Prestoblue dye.
Results:
Five ASCP PDX models were created with 100% initial engraftment rate and >90% engraftment ratio. Immunohistochemistry for p63 (squamous) and mucin components demonstrated the ASCP phenotype. MPseq revealed distinct patterns of aneuploidy and all losses in 17p, 18q and 21q. Each predicted homozygous loss of p16 (CDKN2A) (9p21.3) and heterozygous losses of both TP53 and SMAD4. Two models also predicted double and single losses of PTEN.
Western blot and immunohistochemistry revealed variable MTH1 expression. Cellular spheroids and 2D cultures demonstrated cytostatic sensitivity to the combination of gemcitabine and oxaliplatin as well as the sensitivity to high MTH1 expression tumors and insensitivity to low MTH1 expression tumors. These findings were confirmed with in-vivo treatment studies in tumor bearing PDX models of ASCP.
Conclusion:
ASCP is a rare but more malignant phenotype compared to pancreatic adenocarcinoma. We have generated the world’s first successful models of ASCP and demonstrate variable expression of MTH1. Whole genome sequencing reveals common genomic aberrations. Functional assays using three dimensional organoids demonstrate cytotoxic as well as targeted monotherapy responses. This correlated with therapeutic response in tumor bearing PDX models.
