PREDICTORS OF PANCREATIC IMAGING ABNORMALITIES AND CONCORDANCE BETWEEN ENDOSCOPIC ULTRASOUND AND MAGNETIC RESONANCE IMAGING FOR PANCREATIC CANCER SCREENING IN HIGH-RISK INDIVIDUALS

INTRODUCTION
Pancreatic cancer (PC) screening is recommended in high-risk individuals (HRIs) with familial and genetic risk factors. Endoscopic ultrasound (EUS) and magnetic resonance imaging (MRI) are used in combination as preferred screening modalities. It is unclear if these modalities can be used interchangeably or should be performed simultaneously for maximizing diagnostic accuracy. In a prospective single-institution pancreatic cancer screening registry, we aimed to evaluate the concordance of pancreatic imaging abnormalities in HRIs who underwent paired EUS and MRI and identify predictors of abnormal pancreatic imaging findings on index screening.

METHODS
We identified 100 HRIs with a baseline screening MRI or EUS between July 2021-May 2023. HRIs were defined as individuals eligible for PC screening according to the 2020 AGA Clinical Practice Update. Imaging findings were classified as either normal pancreas (NP) or abnormal pancreatic findings (APF). Continuous variables were summarized as median (IQR) and categorical variables as count (%). Wilcoxon rank sum test, Pearson’s Chi-squared test, and Fisher’s exact test were used to compare clinical and demographic features. Cohen’s kappa was used to evaluate concordance between MRI and EUS.

RESULTS
Solid and cystic pancreatic lesions were identified in 2.1% (4.9%) and 41.5% (48.1%) HRIs on index MRI (EUS), respectively. Among 94 patients with baseline MRI, each decade increase in age was associated with a 2.1-fold increase in the odds of APF (95% CI 1.3-3.4). Patient sex, BMI, ever tobacco use, current alcohol use, family history of PC, and presence of germline mutations were not associated with APF on baseline MRI (Table 1). Only 13/81 (16.0%) patients had NP on EUS, limiting statistical power for group comparisons. Seventy patients had paired index MRI and EUS within 90 days (median time between studies: 2 days; Q1=1, Q3=9, Max=83 days), and were used to evaluate concordance between EUS and MRI. Two patients were found to have a solid mass on both MRI and EUS. Among 68 patients without a pancreatic mass on MRI, a mass was identified on EUS in 2 (2.9%) patients, both low-grade neuroendocrine tumors. For pancreatic cysts, substantial agreement was observed between MRI and EUS (Cohen’s kappa 0.66, 95% CI 0.50-0.82). Three patients (10%) with cysts that were not deemed worrisome on MRI, had worrisome findings on EUS (Table 2).

CONCULSION
In this prospective cohort of HRIs undergoing pancreatic cancer screening, increasing age was the only factor significantly associated with pancreatic imaging abnormalities on index MRI. Pancreatic cystic lesions were frequently identified on index screening, and most were non-worrisome. The agreement between MRI and EUS for pancreatic imaging abnormalities was high, and discordant results included non-worrisome cysts and low-grade neuroendocrine tumors.