2057 - Discrepancies In Pancreatic Cancer Genomic Landscape between the Cancer Genome Atlas and a Large Single-Institution Database

03:00pm - 04:00pm ET

Poster Hall - Exhibit Hall A

Screen: 31

POSTER

Presenter(s)

Susannah Ellsworth, MD - University of Pittsburgh Cancer Centers, Pittsburgh, PA

S. G. Ellsworth¹, A. C. Mueller², A. A. Vera¹, J. Shogan¹, A. H. Zureikat³, J. Y. Zhang⁴, M. Lotze⁵, A. D. Singhi⁶, and H. D. Skinner¹; ¹Department of Radiation Oncology, UPMC Hillman Cancer Center, Pittsburgh, PA, ²Department of Radiation Oncology, UPMC Hillman Cancer Center, University of Pittsburgh, Pittsburgh, PA, ³Department of Surgery, University of Pittsburgh, Pittsburgh, PA, ⁴UPMC, Department of Hematology/Oncology, Pittsburgh, PA, ⁵University of Pittsburgh Hillman Cancer Center, Pittsburgh, PA, ⁶University of Pittsburgh Medical Center, Pittsburgh, PA

Purpose/Objective(s):

Next-generation sequencing is increasingly utilized in assessment of pancreatic ductal adenocarcinoma (PDAC). Emergent data suggests that genomic features may vary by stage in PDAC. The genomic landscape in selected patients receiving radiotherapy (RT) for PDAC is not well characterized. How well public datasets such as the Cancer Genome Atlas (TCGA), which were obtained solely from resected surgical specimens, represent the general population of PDAC patients (pts), is unknown. We compared tumor genomics among pts undergoing RT for PDAC at our institution vs. the Cancer Genome Atlas (TCGA/PanCancer Atlas).

Materials/Methods:

Adult (=18 years) pts who had RT for PDAC were identified from a prospectively maintained institutional database (n=610). Tumor genomics (ThermoFisher OCA Plus) were available in 102 cases. Most analyses (n=54/53%) were performed on surgical specimens; 37% (n=38) from fine-needle biopsy, and 10% (n=10) from metastases or peripheral blood ctDNA. TCGA data was retrieved using cBioPortal.

Results:

Median age in the institutional dataset was 67 years; 47 (46%) were women. Treatment was perioperative in 59 pts (57.8%), definitive in 33 (32.4%), and salvage in 13 (12.7%). KRAS mutation was present in 95 cases (93%); the most common point mutation was KRAS-G12D (n=38/40% of KRAS mutations). Young-onset cancers (=50 years old) accounted for 3/102 cases (3%), none of which were KRAS-wild-type (wt). No microsatellite instability (MSI-high) cases were identified. 2/6 KRAS-wt pts expressed variants of unknown significance in putative driver genes (1 each in FGFR1 and ERBB2). 71 pts (70%) had a TP53 mutation. Mutations in DNA damage repair (DDR) pathways (BRCA1/2, ATM, ATR) were present in 9 cases (8.8%).KRAS mutation was present in 41/43 (95.3%) and 54/58 (93.1%) and TP53 mutation in 31/39 (79.5%) and 40/58 (69.0%) of pts with unresectable vs resectable disease, respectively (p-NS for all comparisons).

In the TCGA PanCancer Atlas, obtained entirely from surgical specimens, 43/184 PDAC pts had RT. Median age was 65; 44% were women. KRAS mutation was present in 26 cases (60%), most commonly at G12D (n=11/42% of KRAS mutations), and TP53 mutation in 22 (51%). Putative driver mutations (BRAF in 2, NTRK1 in 1) were identified in 3/17 KRAS-wt pts in the TCGA atlas; 3 (7.0%) had DDR pathway mutations (1 each of BRCA1, ATM, and ATR). There were no MSI-high pts. Young-onset tumors represented 14.0% of TCGA samples (6/43 patients) and 2 of the KRAS-wt pts.

Conclusion:

Higher frequencies of KRAS and TP53 mutations were observed in a large institutional dataset vs the TCGA PanCancer Atlas. Irradiated pts in TCGA were comparatively enriched for KRAS- and TP53-wt tumors. Frequencies of putative driver mutations, young-onset cancers, and resection status did not fully account for these differences. Caution should be used in extrapolating TCGA genomic landscape data to the general population of PDAC patients treated with RT.