Association between morphological features of necrotizing pancreatitis on endoscopic ultrasound and outcomes of the endoscopic transmural step‐up approach

To investigate the association between necrotic collections on endoscopic ultrasound (EUS) and outcomes of the endoscopic transmural step‐up approach in necrotizing pancreatitis (NP).


| INTRODUCTION
Acute pancreatitis (AP) is a common and potentially lethal disease.
Approximately 10%-20% of patients with AP develop necrotizing pancreatitis (NP). When they develop infected pancreatic necrosis (IPN), the patients present a complex, prolonged clinical course with a mortality of 20%-30%. 1 Over the last decade the management of NP has evolved from open surgery to endoscopic or percutaneous step-up approach due to the efficacy and relatively low morbidity and mortality of these minimally invasive surgical interventions. 2,3 The endoscopic transmural approach is an effective treatment with a mortality of 6% and a complication rate of 36%, and is technically demanding. 4 However, endoscopic factors that may be associated with poor prognosis have been rarely studied.
The extent of necrosis, assessed by contrast-enhanced computed tomography (CECT), is associated with patient prognosis of the endoscopic transmural approach. 5,6 Unfortunately, as acute necrotic collection (ANC) and walled-off necrosis (WON) contain a mixture of fluid and solid necrotic debris, CECT may not distinguish between the solid and liquid components. Endoscopic ultrasound (EUS) can thus be used to facilitate this distinction by detecting the echogenic material in necrotic collections that is suggestive of solid debris. 7 However, its use in predicting the clinical outcomes of endoscopic approach for ANC or WON has rarely been studied. A study has investigated the association between the morphological features of WON on EUS and the outcome of endoscopic transmural drainage (ETD) and revealed that EUS has important therapeutic implications with collections of a large size and more solid debris requiring aggressive therapy to achieve successful outcomes. 8 However, the study had a relatively small sample size (n = 43), and no multivariate analysis was performed; therefore, the results might have been affected by the confounding factors. Thus, further research on the value of EUS for predicting the clinical outcomes of the endoscopic transmural approach for ANC or WON is warranted.
In the current study, we aimed to retrospectively analyze the patients with ANC or WON who underwent ETD as the initial intervention and to investigate the relationship between the morphological features of pancreatic necrotic collection on EUS and the clinical outcomes of the endoscopic transmural step-up approach. Furthermore, we aimed to identify risk factors for predicting clinical failure of the endoscopic transmural step-up approach for managing NP and to develop a novel nomogram to predict this failure.

| Study design and patient enrollment
This was a single-center, retrospective cohort study. Adult patients with NP who underwent ETD or endoscopic transmural necrosectomy (ETN) at the Department of Gastroenterology of the First Affiliated Hospital of Nanchang University, a tertiary care referral hospital in Nanchang (Jiangxi Province, China), from April 2015 to April 2020 were recruited. The exclusion criteria were as follows: (a) patients who underwent ETD without EUS guidance (due to the lack of images); (b) patients whose only collection was drained by both endoscopic and percutaneous approaches, as including them might have induced a confounding effect on the research results; and (c) those who were lost to the follow-up of at least 6 months after their discharge. Computed tomography (CT) scan was repeated according to the patient's condition until the necrotic collections had almost completely disappeared, with no local complications. This study was approved by the Ethics Committee of the First Affiliated Hospital of Nanchang University (no. 2011001). Written informed consent was waived due to the retrospective nature of this study.

| EUS and data collection
EUS images were continuously collected at the time of the initial ETD and were retrieved from the endoscopic system, which included more than 30 images or a video, for a thorough evaluation of each collection from different angles. To assess the interobserver and intra-observer agreement, the retrieved EUS images or videos were reviewed by two experienced endoscopists who were blinded to the clinical outcomes of the patients. The κ statistics were used to assess the interobserver and intra-observer agreements during the evaluation of the amount of solid necrotic debris; a κ statistic of 1 indicated excellent interobserver and intra-observer agreements. The detailed morphological features of the necrotic collection assessed by EUS, including its location and size, were reviewed with a focus on the amount of solid necrotic debris. Echogenic material presenting in the necrotic collection was suggestive of solid debris. All EUS images were evaluated to determine the amount of solid debris, and the image section with the most solid necrotic debris was chosen for quantification. The chosen image was divided into four quadrants and the amount of solid necrotic debris was quantified as a percentage of the total collection size. According to the amount of solid necrotic debris, the patients were divided into three groups: group 1, <30%; group 2, 30%-50%; and group 3, >50% ( Figure 1). These cut-off values were chosen according to the Balthazar CT severity index.
Baseline characteristics (age, sex, body mass index [BMI], etiology of AP, status of smoking and alcohol consumption, disease severity), interventions and clinical outcomes of the patients were obtained from a prospectively maintained medical database, which is a data repository of all clinical data obtained from AP patients admitted to our department. These data were extracted and recorded by a special research assistant. Their body temperature, levels of C-reactive protein (CRP) and procalcitonin, systemic inflammatory response syndrome (SIRS) score and presence of organ failure 1 week before the initial ETD were also collected. Additionally, organ failure events, including respiratory, circulatory or renal failure, were also recorded, as indicated by a score of 2 or higher using the modified Marshall scoring system. 7

| Interventions
All management decisions on the interventions were made by a multidisciplinary team of endoscopists, pancreatologists, surgeons, interventional radiologists, radiologists and intensivists using our previously described algorithm. 10,11 The indications for NP inter- F I G U R E 1 Association between the morphological features of necrotic collections in endoscopic ultrasound (EUS) and endoscopic transmural necrosectomy (ETN) sessions. Necrotic collection with A, predominantly liquid content and <30% solid necrotic debris successfully treated in a single session of ETN; B, with 30%-50% and C, >50% solid necrotic debris that required two sessions of ETN; D, with predominantly purulent liquid content and minimal necrosis also required a single session of ETN

| Baseline characteristics of the patients
A total of 140 patients with NP were managed by ETD or ETN at our center during the study period. Four patients were excluded because of the lack of EUS images (n = 2) or loss to follow-up (n = 2). Another two patients whose primary collection was drained by both endoscopic and percutaneous approaches were also excluded. Finally, 134 patients were included in our cohort. The study flowchart is shown in Figure 2. Most necrotic collections (79.9%) located at the body and/or tail of the pancreas, with 6.0% located at the head and the other 14.2% located at the whole pancreas. The median necrotic collection size was 8.5 cm (IQR 7.0-10.0 cm), and the median extent of solid necrotic debris was 30% (IQR 20%-50%; range 10%-90%). A total of 52, 59 and 23 patients had solid necrotic components of <30% (group 1), 30%-50% (group 2) and >50% (group 3) of the total size of the collection, respectively. There were no significant differences among the three groups in terms of patients' sex, age, BMI, etiology of AP, status of smoking and alcohol consumption, and disease severity (assessed by body temperature, C-reactive protein and procalcitonin levels, SIRS score, and the presence of organ failure 1 week before the initial ETD. Baseline characteristics of the patients are summarized in Table 1.

| Interventions
In our cohort, infection was the primary indication for intervention in over half the patients, and was significantly more common in patients with larger amounts of solid necrotic debris than those with    Table S1.

| Clinical outcomes
The univariate logistic regression analysis for the identification of factors for clinical failure in the endoscopic transmural step-up approach for ANC or WON was performed (Table 3). Factors associated with clinical failure in the unadjusted univariate models (P < 0.1) were further included in the multivariate models. The results showed that procalcitonin (adjusted odds ratio [aOR] 6.14, 95% confidence interval [CI] 1.40-26.94, P = 0.016) and any organ failure (aOR 11.51, 95% CI 2.42-54.78, P < 0.01) were found to be independently associated with clinical failure (Table 4). Based on the multivariate analysis, T A B L E 1 Baseline characteristics of patients with necrotizing pancreatitis who underwent endoscopic transmural drainage (ETD) compared between groups determined by the amount of solid necrotic debris on endoscopic ultrasound (EUS) a All (n = 134) Group 1 (n = 52) Group 2 (n = 59) Group 3 (n = 23) P value Location of collection, a n (%) 0.082   of endoscopic procedures required for a successful outcome. 8 Although this was a promising finding, our study included a relatively small sample and the results might have been affected by the confounding factors. Our results are consistent with the previous study, showing that patients with more solid necrotic debris required more ETN procedures to achieve clinical success. However, the correlation coefficient was low (r = 0.556) and the extent of necrosis was not sig- In our study, procalcitonin and any organ failure were found to be independent risk factors for predicting clinical failure of the endoscopic transmural approach. Procalcitonin is an early rapid-response marker of SIRS, which correlates well with the incidences of infected necrosis, organ failure and death in patients with NP. 12 Our results suggest that serum procalcitonin level of ≥0.5 ng/mL 1 week before the initial ETD, compared with that of less than 0.5 ng/mL, could serve as a risk factor to assess patients' response to the endoscopic transmural approach. The OR of any organ failure 1 week before the initial ETD was almost 12, illustrating its value in predicting clinical failure. A relatively higher risk of mortality and disease severity may account for a higher risk of clinical failure in patients who develop organ failure-related complications. 13 Similarly, a post hoc analysis revealed that male sex, multiple organ failure, an increasing percentage of pancreatic necrosis and heterogeneity of the collections are negative predictors for the success of catheter drainage in infected necrotizing pancreatitis. 6 However, this study included only 17 patients treated via endoscopy, and the features of necrosis were assessed on CECT, on which it is difficult to distinguish solid component from fluid component. 14 It is important to highlight the wide variability in the use of ETD or ETN. We adopted a strategy based on the endoscopically centered stepup approach. First, ETD was found to be an optimal treatment for primary necrotic collections adjacent to the stomach or duodenum, with the subsequent use of ETN when required. Although we did not adopt methods such as multiple stent placement, aggressive irrigation, or multiple transluminal gateway techniques to improve the efficiency of the drainage, the prognosis turned out to be good. The all-cause mortality rate was 10.4%, which was somewhat lower than that of 15%-39% reported in the literature and was similar to that reported in recent randomized trials on the endoscopic step-up approach. 2,3,[15][16][17] We believe that the low mortality rate in our study might be due to the following reasons: (a) our emphasis on a multidisciplinary team approach and the use of a minimally invasive interventional strategy as the first treatment choice; (b) our choice of metal stents provided a larger tract for drainage; and (c) the delay in performing endoscopic necrosectomy for at least 4 weeks. 10,11 The potential strength of our study is that it was a cohort study based on a prospectively maintained database including all patients with NP undergoing ETD during the study period. Thus, the study sample is representative of the entire spectrum of patients rather than a preselected group of patients, thereby reducing selection bias.
F I G U R E 3 A, A nomogram based on the multivariate logistic regression analysis was developed for the prediction of clinical failure. B, The receiver operating characteristic curve and C, the decision curve analysis showing good discrimination, calibration, and clinical utility of the novel prediction model. Abbreviyation: AUROC, area under the receiver operating characteristic curve; PCT, procalcitonin. , model; , all; , none However, there were some limitations to the current study.
First, although two experienced endoscopists reassessed the EUS images and quantification of the solid debris was performed, one section of an EUS image may not represent the total amount of solid necrotic debris. More objective measurements of the volume of necrosis should be explored in future studies. Second, this was a single-center study, and all procedures were performed at a highly specialized tertiary care center by experienced clinicians in pancreatic endotherapy, which may not be available at small institutions.
Thus, the representativeness of the conclusions might have been affected.