Subscribe to RSS
DOI: 10.1055/s-0040-1704172
IMPACT OF CYST FLUID GNAS AND KRAS MUTATIONAL ANALYSIS ON PANCREATIC CYST CATEGORIZATION IN PATIENTS EVALUATED BY EUS-FNA: A SINGLE-CENTER EXPERIENCE OF 103 PATIENTS
Publication History
Publication Date:
23 April 2020 (online)
Aims Preoperative categorization of pancreatic cysts (PC) is hampered by lack of accurate imaging and ancillary markers. Somatic mutations in KRAS and GNAS in cystic fluid are specific for mucinous PC (mPC).
Aims: To describe the frequency of GNAS/KRAS mutations in PC fluid obtained by endoscopic ultrasound-fine-needle aspiration (EUS-FNA) and to analyze its impact over imaging/CEA/cytology in predicting mPC.
Methods Patients with PC who underwent USE-PAAF and GNAS/KRAS analysis in cystic fluid were prospectively collected. PC were categorized as mPC, non-mPC or indeterminate (iPC) based on magnetic resonance cholangiopancreatography and USE, intracystic carcinoembrionic antigen concentration (CEA) and fluid cytology.
Results 103 patients were analyzed. CEA was performed in 72 (70%) and cytology in 84 (81%). Imaging/CEA/cytology identified 47 (46%) mPC, 4 (4%) non-mPC (3 pseudocysts, 1 paraduodenal cyst) and 52 (50%) iPC. Either GNAS or KRAS mutations were found in 59 (57%) PC (38 mPC, 0 non-mPC, 21 iPC). Hence, GNAS/KRAS mutations reclassified 41% of iPC into mPC, yielding 69 (70%) mPC and 31 (30%) iPC. Of 69 mPC, 59 (85%) were GNAS and/or KRAS mutated, 49 (47%) only GNAS, 28 (27%) only KRAS and 19 (27%) both GNAS and KRAS. Mutational analysis showed better sensibility and specificity (85% and 100%) than CEA (35% and 75%) and cytology (33% and 96%) to predict mPC. Within CEA negative (< 192 ng/mL) mPC (65%), 90% showed GNAS and/or KRAS mutations. Of 16 PC that were surgically resected, 12/13 (93%) mPC had GNAS/KRAS mutations and 3/3 non-mPC showed no mutations.
Conclusions There is a high prevalence of GNAS/KRAS mutation in mPC fluid obtained by EUS-FNA, which increases the diagnostic yield over imaging/CEA/cytology. These results support its use in the decision tree algorithm for PC typecasting.
#