Tu1939 A Retrospective Single Center Analysis of Survival and Prognostic Factors of Adjuvant Gemcitabine Following Curative-Intended Surgery of Pancreatic Adenocarcinoma

Abstract

G A A b st ra ct s were de-identified and evaluated by 4 blinded CyPs (3 experienced CyPs, 1 fellow). A modified validated scoring system was used to assess specimen quantity and quality and standardized among all CyPs (Table 1). Final cytologic diagnosis was categorized as: insufficient, benign, atypical, suspicious for malignancy or malignant and final clinical diagnosis as benign or malignant. IOV was calculated using multi-rater kappa (κ) statistics with 95% CI. Bivariate analyses were performed comparing cases with and without uniform agreement followed by logistic regression with backward selection to model the likelihood of uniform agreement. Assuming an overall κ of 0.8 amongst 4 CyPs, prevalence of malignancy of 0.6, and lower limit of 95% CI ≥ 0.7, a sample size of at least 72 cases was determined. RESULTS: 99 cases were included [49% males; mean age 64 yrs] of which 44% presented with weight loss and 28% jaundice. On EUS, mean lesion size was 25 mm, with distribution of lesions being: head/uncinate 56%, neck 8%, body 21% and tail 15%. EUS findings consistent with chronic pancreatitis were noted in 10% of cases. Final cytologic diagnosis revealed malignant diagnosis in 60% of cases, suspicious 4%, atypical 5%, benign 28% and inadequate 3%. IOV for overall final cytologic diagnosis was moderate (κ =0.45) with minimal improvement when combining suspicious + malignant diagnoses (κ=0.54). IOV was slight to fair (κ 0.040.32) for individual parameters (Table 1). Presenting symptom of jaundice (p<0.001), lesion size (p=0.04), number of nucleated/diagnostic cells (p<0.001), degree of GI contamination (p=0.03) and final cytologic (p=0.02) and clinical diagnosis (p=0.003) were significant factors on bivariate analysis (Table 2). On multivariate analysis the predictor for uniform agreement among CyPs was final clinical diagnosis of malignancy [OR 3.99 (CI 1.52-10.49)].CONCLUSIONS: Results of this study demonstrate significant IOV among CyPs for the final cytologic diagnosis and individual quantity and quality measures in the assessment of pancreatic EUSFNA specimens. The final clinical diagnosis of malignancy is the strongest predictor for agreement among CyP. Confirming these results and refinement of EUS-FNA scoring criteria in future multicenter studies is imperative. Table 1: Standardized scoring tool to assess individual EUS-FNA passes and final cytologic diagnosis and IOV of individual parameters