Pearson, Randall K.

Objective The diagnosis of autoimmune pancreatitis (AIP) is challenging. Sonographic and cross-sectional imaging findings of AIP closely mimic pancreatic ductal adenocarcinoma (PDAC) and techniques for tissue sampling of AIP are suboptimal. These limitations often result in delayed or failed diagnosis, which negatively impact patient management and outcomes. This study aimed to create an endoscopic ultrasound (EUS)-based convolutional neural network (CNN) model trained to differentiate AIP from PDAC, chronic pancreatitis (CP) and normal pancreas (NP), with sufficient performance to analyse EUS video in real time. Design A database of still image and video data obtained from EUS examinations of cases of AIP, PDAC, CP and NP was used to develop a CNN. Occlusion heatmap analysis was used to identify sonographic features the CNN valued when differentiating AIP from PDAC. Results From 583 patients (146 AIP, 292 PDAC, 72 CP and 73 NP), a total of 1 174 461 unique EUS images were extracted. For video data, the CNN processed 955 EUS frames per second and was: 99% sensitive, 98% specific for distinguishing AIP from NP; 94% sensitive, 71% specific for distinguishing AIP from CP; 90% sensitive, 93% specific for distinguishing AIP from PDAC; and 90% sensitive, 85% specific for distinguishing AIP from all studied conditions (ie, PDAC, CP and NP). Conclusion The developed EUS-CNN model accurately differentiated AIP from PDAC and benign pancreatic conditions, thereby offering the capability of earlier and more accurate diagnosis. Use of this model offers the potential for more timely and appropriate patient care and improved outcome.

OBJECTIVES: Expression of prostaglandin biosynthetic pathway enzymes in mucinous pancreatic cysts is unknown. Cyclooxygenase-2 (COX-2) inhibition is a potential cancer chemoprevention strategy for these lesions. We evaluated the expression of COX-2, cytosolic phospholipase A2 (cPLA2), and protein kinase B (AKT) in the epithelium of pancreatic cysts and correlated enzyme expression with aspirin (ASA) use and cyst fluid prostaglandin E 2 (PGE 2 ) concentration. METHODS: Pathology of 80 resected pancreatic cysts was reviewed. Expression of COX-2, cPLA2, and AKT was quantified by tissue immunohistochemistry immunoreactivity scores (IRSs). IRS values were compared between cyst types and (in 30 cases) with matched cyst fluid PGE 2 concentrations. RESULTS: The mean IRS was higher in the epithelium of mucinous vs nonmucinous cysts for COX-2 (6.1 ± 4.7 vs 3.2 ± 2.8, P = 0.01) and cPLA2 (6.9 ± 3.0 vs 2.9 ± 2.9, P < 0.001). Cyst epithelial COX-2 expression was higher in mucinous cysts with low-grade dysplasia vs those with high-grade dysplasia or invasive carcinoma (IRS 8.0 ± 3.9 vs 1.5 ± 2.9, P < 0.001), whereas the opposite was found for cPLA2 (6.2 ± 3.0 vs 8.6 ± 2.3, P = 0.005). Cyst fluid PGE 2 concentrations did not correlate with either the IRS or a history of low- to moderate-dose ASA use. CONCLUSIONS: COX-2 and cPLA2 are overexpressed in the epithelium of mucinous pancreatic cysts. COX-2 and/or cPLA2 inhibition might prevent the emergence or progression of mucinous pancreatic cysts, but higher doses of ASA or nonsteroidal anti-inflammatory drugs may be necessary to substantially inhibit cyst epithelial COX-2 activity.

BACKGROUND: There is lack of consensus on post-operative surveillance for resected non-invasive intraductal papillary neoplasms (IPMNs). In this study we explored risk factors for subsequent PC in patients with MD-IPMN undergoing partial pancreatectomy. METHODS: We searched the Mayo Clinic surgical pathology database for all cases of resected MD-IPMN between 1997 and 2014. Cases with histologically confirmed main pancreatic duct involvement either isolated or in a mixed pattern with branch-duct involvement were included. Outcomes of PC in the remnant pancreas, and death related to MD-IPMN were assessed with survival analyses (Kaplan–Meier and Cox regression). RESULTS: Among the 179 patients with resected MD-IPMN the incidence of concomitant PC and high-grade dysplasia (HGD) in the resected specimen was 23 and 14%, respectively. The mean duration of follow-up was 4.31 years (range 0.12–13.5 years). Excluding 28 subjects who either underwent initial total pancreatectomy or partial pancreatectomy with surgical margins positive for PC/HGD, the 5-year incidence of subsequent PC was 12%, including 60.6% and 15.6% in those with initial PC and HGD, respectively. The 10-year incidence of PC was 21.2% overall, 60.6% for PC, 38.3% for HGD, and 3.0% for LGD. Risk of subsequent PC was significantly higher for those with initial PC compared with HGD (HR = 4.95, 95% CI: 1.63–15.03, p = 0.005 and for HGD compared with LGD (HR = 11.30, 95% CI: 1.55–82.26, p = 0.017). CONCLUSIONS: Patients with MD-IPMN with PC or HGD undergoing segmental pancreatectomy are at higher risk of subsequent PC and may benefit from post-operative surveillance. The post-operative surveillance intervals in resected MD-IPMNs need to be tailored based on dysplasia grade.

Human gastric smooth muscle tumors (leiomyosarcomas) have been shown to express cholecystokinin (CCK) binding sites that are functionally similar to physiologically important receptors present on their cells of origin. In this work, we have applied affinity-labeling techniques using 125I-D-Tyr-Gly-[Nle28,31]CCK-(26-33) to attempt to define the ligand-binding subunit of this receptor, and we have used the receptor antagonist L364,718 and deglycosylating enzymes to compare this molecule with well-defined CCK receptors on the classical peripheral targets (pancreas and gallbladder) of this hormone. To validate the use of 125I-D-Tyr-Gly-[Nle28,31]CCK-(26-33) for this tissue, we demonstrated that it bound to leiomyosarcoma membranes in a rapid, reversible, saturable, specific, and high-affinity manner (Kd = 0.8 nM). Although previous affinity labeling of this tissue with a CCK-33-based probe identified multiple bands, only one of those candidate proteins was predominantly labeled in the present work (Mr 100,000) by using a probe that is cross-linked through a site in greater proximity to the receptor-binding domain. Labeling was inhibited in a concentration-dependent manner by CCK-8 but not by structurally unrelated ligands. Although endo-beta-N-acetyl-glucosaminidase F digestion shifted this band by Mr 5,000, demonstrating that it was a glycoprotein, the deglycosylation product was very different from other CCK receptors studied. Also, unlike pancreatic and gallbladder CCK receptors, affinity labeling of this receptor was not affected by L364,718. These observations confirm that the gastric smooth muscle tumor CCK receptor represents a receptor subtype that is distinct from other peripheral CCK receptors, biochemically as well as functionally.

Objective Fukuoka consensus guidelines classify pancreatic cystic lesions (PCLs) presumed to be intraductal papillary mucinous neoplasms (IPMNs) into Fukuoka positive (FP) (subgroups of high-risk (HR) and worrisome features (WFs)) and Fukuoka negative (FN) (non-HR feature/WF cysts). We retrospectively estimated 5-year risk of pancreatic cancer (PC) in FN, WF and HR cysts of patients with PCL-IPMN. Design From Mayo Clinic databases, we randomly selected 2000 patients reported to have a PCL; we excluded inflammatory or suspected non-IPMN cysts and those without imaging follow-up. We re-reviewed cross-sectional imaging and abstracted clinical and follow-up data on PCL-IPMNs. The study contained 802 patients with FN cysts and 358 with FP cysts. Results Patients with PCL-IPMN had median (IQR) follow-up of 4.2 (1.8–7.1) years. Among FN cysts, 5-year PC risk was low (2–3%) regardless of cyst size (p=0.67). After excluding events in the first 6 months, 5-year PC risk remained low (0–2%) regardless of cyst size (p=0.61). Among FP cysts, HR cysts (n=66) had greater 5-year PC risk than WF cysts (n=292) (49.7% vs 4.1%; p<0.001). In HR cysts, 3-year PC risk was greatest for obstructive jaundice versus enhancing solid component or main pancreatic duct >10 mm (79.8% vs 37.3% vs 39.4%, respectively; p=0.01). Conclusions Fukuoka guidelines accurately stratify PCL-IPMNs for PC risk, with FN cysts having lowest and HR cysts having greatest risk. After 6-month follow-up, WF and FN cysts had a low 5-year PC risk. Surveillance strategies should be tailored appropriately.

Objective Idiopathic duct-centric chronic pancreatitis (IDCP), also known as type 2 autoimmune pancreatitis (AIP), is an uncommon subtype of AIP. International Consensus Diagnostic Criteria for IDCP propose that the diagnosis requires pancreatic histology and/or concurrent IBD. We examined our experience with IDCP (type 2 AIP) to assess the appropriateness of these criteria, and identify unique characteristics in patients presenting with acute pancreatitis. Design We reviewed the Mayo Clinic AIP database through May 2014 to identify subjects with either definitive (n=31) or probable (n=12) IDCP. We compared demographic and clinical factors based on strength of diagnostic confidence (definitive versus probable), presence of IBD, and acute pancreatitis as the presenting manifestation. Relapse-free survival was determined using the Kaplan-Meier method. Results The clinical profiles were similar irrespective of the diagnostic criteria fulfilled. Common clinical presentations included acute pancreatitis (n=25, 58.1%, 12 of whom (27.9%) had recurrent pancreatitis) and pancreatic mass/obstructive jaundice (n=15, 34.9%). The cumulative relapse rate was 10.6% at 3 years (median follow-up 2.9 years). Relapse-free survival was similar for the different diagnostic categories, but was decreased in those initially presenting with acute pancreatitis (p=0.047) or treated with steroids (vs surgery, p=0.049). Conclusions The current diagnostic classification of probable IDCP and the inclusion of IBD as a supportive criterion appear valid, because patients have similar clinical profiles and disease-related outcomes to those with definitive IDCP. Concurrent IBD, especially in young patients, may suggest when IDCP is the underlying cause of recurrent acute pancreatitis, but additional studies are needed for validation.