Pancreatic cystic neoplasms

Pancreatic cystic neoplasms: Adherence to management recommendations and associated endosonographic cyst characteristics

BACKGROUND

The purpose of this study is to assess patients' compliance to recommendations after evaluation of pancreatic cystic neoplasm (PCN) with EUS and investigate how the presence of "worrisome" characteristics including cyst's size, main pancreatic duct dilation and presence of mural nodules might affect the adherence rates of management recommendations.

METHODS

We performed a retrospective cohort study of patients at a private tertiary hospital who were referred for evaluation with EUS over a 5-year period (2015-2019), after the diagnosis of pancreatic cystic neoplasm during radiological imaging.

RESULTS

We included 111 patients (mean age 64.1 years, SD = 13.9) with PCN. After the EUS examination, 16 patients were referred for surgical resection, 4 patients needed no further follow up and 91 patients were recommended to follow imaging surveillance. In total, 70 (63.1%) subjects adhered to surveillance recommendations. In the group of subjects who adhered to surveillance, cyst size ≥3cm was found in 27 (38.6%) patients, main pancreatic duct diameter ≥5mm in 12 (17.1%) subjects and only 3 (4.3%) pancreatic cysts demonstrated mural nodules. However, none of the aforementioned cystic "worrisome features" was significantly correlated with increased adherence to follow up (p = 0.709, P = 0.642 and P = 0.630, respectively).

CONCLUSIONS

Although the majority of patients with PCN adhered to given recommendations after EUS examination, the number of noncompliant subjects was noticeable. The presence of cystic "worrisome" features did not correlate with an increased compliance rate to suggested management plan. Further prospective studies are needed to elucidate the factors that may enhance patients' adherence.

Use of Artificial Intelligence Deep Learning to Determine the Malignant Potential of Pancreatic Cystic Neoplasms With Preoperative Computed Tomography Imaging

BACKGROUND

Society consensus guidelines are commonly used to guide management of pancreatic cystic neoplasms (PCNs). However, downsides of these guidelines include unnecessary surgery and missed malignancy. The aim of this study was to use computed tomography (CT)-guided deep learning techniques to predict malignancy of PCNs.

MATERIALS AND METHODS

Patients with PCNs who underwent resection were retrospectively reviewed. Axial images of the mucinous cystic neoplasms were collected and based on final pathology were assigned a binary outcome of advanced neoplasia or benign. Advanced neoplasia was defined as adenocarcinoma or intraductal papillary mucinous neoplasm with high-grade dysplasia. A convolutional neural network (CNN) deep learning model was trained on 66% of images, and this trained model was used to test 33% of images. Predictions from the deep learning model were compared to Fukuoka guidelines.

RESULTS

Twenty-seven patients met the inclusion criteria, with 18 used for training and 9 for model testing. The trained deep learning model correctly predicted 3 of 3 malignant lesions and 5 of 6 benign lesions. Fukuoka guidelines correctly classified 2 of 3 malignant lesions as high risk and 4 of 6 benign lesions as worrisome. Following deep learning model predictions would have avoided 1 missed malignancy and 1 unnecessary operation.

DISCUSSION

In this pilot study, a deep learning model correctly classified 8 of 9 PCNs and performed better than consensus guidelines. Deep learning can be used to predict malignancy of PCNs; however, further model improvements are necessary before clinical use.

64-Slice spiral computed tomography and three-dimensional reconstruction in the diagnosis of cystic pancreatic tumors

The present study aimed to describe the characteristics of cystic pancreatic tumors using computed tomography (CT) and to evaluate the diagnostic accuracy (DA) of post-imaging three-dimensional (3D) reconstruction. Clinical and imaging data, including multi-slice spiral CT scans, enhanced scans and multi-faceted reconstruction, from 30 patients with pathologically confirmed cystic pancreatic tumors diagnosed at the Linyi People's Hospital between August 2008 and June 2014 were retrospectively analyzed. Following the injection of Ultravist® 300 contrast agent, arterial, portal venous and parenchymal phase scans were obtained at 28, 60 and 150 sec, respectively, and 3D reconstructions of the CT images were generated. The average age of the patients was 38.4 years (range, 16-77 years), and the cohort included 5 males and 25 females (ratio, 1:5). The patients included 8 cases of mucinous cystadenoma (DA), 80%]; 9 cases of cystadenocarcinoma (DA, 84%); 6 cases of serous cystadenoma (DA, 100%); 3 cases of solid pseudopapillary tumor (DA, 100%); and 4 cases of intraductal papillary mucinous neoplasm (DA, 100%). 3D reconstructions of CT images were generated and, in the 4 cases of intraductal papillary mucinous neoplasm, the tumor was connected to the main pancreatic duct and multiple mural nodules were detected in one of these cases. The DA of the 3D-reconstructed images of cystic pancreatic tumors was 89.3%. The 64-slice spiral CT and 3D-reconstructed CT images facilitated the visualization of cystic pancreatic tumor characteristics, in particular the connections between the tumor and the main pancreatic duct. In conclusion, the 3D reconstruction of multi-slice CT data may provide an important source of information for the surgical team, in combination with the available clinical data.