Impact of preoperative endoscopic procedures on adverse event rates after surgical resection for main-duct and mixed-type intraductal papillary mucinous neoplasms (IPMNs)

Abstract

Background:

Main-duct (MD-) and mixed-type (MT-) IPMNs harbor an increased risk of pancreatic cancer and warrant surgical resection. Preoperative endoscopic retrograde cholangiopancreatography (ERCP) and endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) are important in the diagnosis of IPMNs. The aim of this study was to investigate whether endoscopic procedures manipulating the MD impact postoperative adverse events in patients with MD- and MT-IPMNs.

Methods:

We performed a retrospective study of 369 patients who underwent resections for MD- or MT-IPMN at two tertiary centers (2000–2019). Multivariable logistic regression analyses were performed for postoperative adverse events to compare the risks between intervention (ERCP, EUS-FNA with branch duct (BD) aspirated, EUS-FNA with MD aspirated from the duct directly or cyst/mass arising from MD) versus no-intervention group.

Results:

33.1 % of patients had a preoperative ERCP and 69.4 % had EUS-FNA. Postoperative adverse events included: 30-day readmission (12.7 %), delayed gastric emptying (13.8 %), pancreatic fistula (10.3 %), abdominal abscess (5.7 %), cardiopulmonary adverse events (11.4 %), and mortality (1.4 %). The model was adjusted for potential confounders. There were no significant differences between the ERCP and no-ERCP groups for specific adverse events. Compared to no-EUS-FNA groups, groups of EUS-FNA with BD aspiration and EUS-FNA with MD aspiration from the main pancreatic duct directly or cyst/mass arising from MD did not show a significant increase in specific adverse events.

Conclusions:

Postoperative adverse events were not significantly increased among patients who had ERCP or EUS-FNA before surgical resection for MD- or MT-IPMNs. Endoscopic procedures directly sampling the MD can be safely pursued for diagnostic purposes in selected cases.

1. Introduction

Intraductal papillary mucinous neoplasms (IPMNs) are a type of pancreatic cystic lesions that arise from the pancreatic ducts and produce mucin. They are of particular importance since they are precursors to pancreatic cancer, the fourth leading cause of cancer death in the United States [1]. Based on their anatomic involvement, IPMNs can be categorized into main-duct IPMNs (MD-IPMNs) predominantly involving the epithelium of the main pancreatic duct, branch-duct IPMNs (BD-IPMNs) involving only the side branches, and mixed-type IPMNs (MT-IPMNs) involving both the main pancreatic duct and side branches.

Several guidelines exist for the management of patients with IPMNs [2–6], with surgery recommended in patients with high-risk stigmata including jaundice, main pancreatic duct of greater than or equal to 10 mm or an enhancing mural nodule [6]. Additionally, MD-IPMNs can have diffuse or segmental dilation of the main duct (MD), and research suggests that maximal MD diameter, extent of segmental dilation, and diameter of the non-segmental MD are associated with risks of high-grade dysplasia or adenocarcinoma [7]. While increasing main ductal dilation could confer risks of cancer, in clinical practice it is not always clear that the MD dilation is due to an IPMN. In MT-IPMNs, the dilation of the main pancreatic duct from 5 to 9 mm is considered a worrisome feature [5] and it is not clear if this dilation is due to involvement of the main duct or from the backflow of mucin into the main pancreatic duct.

Endoscopic ultrasound (EUS) with fine-needle aspiration (FNA) is often used to confirm the diagnosis of a mucinous cyst, and/or to further classify the IPMN if there are one or more worrisome features on cross-sectional imaging. It provides high-resolution imaging of the pancreas to further characterize the cystic lesion and mural nodules, and allows sampling of the cystic fluid and mural nodules for cytological evaluation, biochemistry and genetic testing for risk stratification. Additionally, endoscopic retrograde cholangiopancreatography (ERCP) is often used preoperatively to relieve biliary or pancreatic duct obstruction with stent placement, obtain cytology by aspiration of mucinous fluid or biopsy, and perform pancreatoscopy. While these procedures are relatively safe with a low complication rate, FNA of the main pancreatic duct is rarely performed given the risk of pancreatitis [8]. This risk is not clearly established.

We performed a bi-center retrospective review of all patients with MD-IPMNs and MT-IPMNs who underwent surgical resection at our institutions and compared patients who underwent preoperative ERCP and/or EUS-FNA with those who did not. Our goal was to examine the role of main pancreatic duct manipulation through EUS-FNA or ERCP and determine if there was any increase in surgical adverse events.

2. Methods

2.1. Study design

We performed a retrospective study of prospectively maintained database of patients who underwent surgical resection for IPMNs between 2000 and 2019 at Massachusetts General Hospital (MGH) and Johns Hopkins Hospital (JHH), two tertiary care centers in the United States. The databases documented patients who underwent surgical resection for MD-, MT-IPMNs (MGH, JHH) and BD-IPMNs (MGH). There were 305 patients identified from the MGH hospital system. The inclusion criteria were all pathologically confirmed MD- and MT-IPMNs undergoing surgical resection. On further chart review, 59 patients were screened out of the study. 4 patients were excluded because they were found to not have IPMN on final pathology, 45 patients were excluded because they were found not to have main duct involvement on final pathology, and 10 patients were excluded because they did not undergo ERCP and had unknown status for EUS-FNA. There were 123 patients identified in the JHH database, and none of them were excluded based on the criteria. The full inclusion and exclusion criteria are described in Fig. 1. The informed consent from patients to be included in the databases of the study was obtained via in person discussion with physician. The protocol to access the data was approved by the institutional review boards at MGH and JHH.

Fig. 1.

Patient Selection and Inclusion using databases from Massachusetts General Hospital (MGH) and Johns Hopkins Hospital (JHH).

2.2. Endoscopic procedures

All endoscopic procedures were performed by well-trained endoscopists. At MGH, EUS and ERCP were performed under monitored anesthesia care (MAC) or general anesthesia. The type of anesthesia was determined by anesthesiologists based on patients’ medical needs. At JHH, all endoscopic procedures were performed under MAC.

The indication for EUS was evaluation of a pancreatic cyst or mass. In EUS, the endoscope was introduced through the mouth and advanced to the second portion of the duodenum under direct visualization. Water was used as necessary to provide an acoustic interface. Imaging was obtained at 5 and 7.5 MHz. At MGH, intravenous prophylactic antibiotics were given to all patients per protocol for EUS, while at JHH, it was given if FNA was performed and patients were also instructed to take three days of ciprofloxacin after returning home.

Indications for ERCP included recurrent acute pancreatitis (n = 22), dilated main pancreatic duct (n = 56), further evaluation of abdominal pain (n = 6), placement or exchange of stents (n = 17), evaluation of pancreatic cyst or mass (n = 7), and concurrent choledocholithiasis (n = 1). The indication was unknown for 13 patients. Among patients who received ERCP, 54 patients had stent placement, including 24 patients with pancreatic stents, 12 with biliary stents, 3 with both, and 15 with stents of unknown location. 18 patients had biliary or pancreatic sphincterotomy, 6 had biopsy of pancreatic cyst or mass, 5 had pancreatoscopy, 3 had pancreatic duct brushing, 3 had pancreatic duct dilation, 2 had aspiration of pancreatic fluid, 2 had sphincteroplasty, and 1 had gallstone extraction. Sphincterotomy was performed based on patients’ medical needs at MGH. Rectal indomethacin was given to all patients following ERCP based on clinical guidelines [9]. At JHH, sphincterotomy was performed in patients who did not have previous evidence of sphincterotomy. A pancreatogram was obtained during ERCP. Pancreatic stenting and rectal indomethacin would be given based on clinical guidelines [9]. Spyglass pancreatoscope was used in cases of dilated MD-IPMN. No patients received throughthe-needle confocal endomicroscopy, through-the-needle biopsy, or cystoscopy in either cohort.

2.3. Baseline characteristics

The following baseline characteristics were collected and included in the study: age, sex, ethnicity, body mass index (BMI), smoking status, prior pancreatic or biliary surgery, American Society of Anesthesiologists (ASA) class, type of IPMN (MD- or MT-IPMN), epithelial subtype, histopathology grade, IPMN size on imaging, type of pancreas surgery, pathological margin, large vessel involvement, lymphovascular or perineural invasion, endoscopic procedure (ERCP, EUS-FNA, none), post-ERCP or post-EUS adverse events, time interval between the last endoscopic procedure (EUS or ERCP that occurred last) and surgery.

2.4. Outcomes

Outcome analysis was performed on patients with pathologically confirmed MD- or MT-IPMN. Patients were divided into intervention and no-intervention groups. The intervention group included patients who had ERCP, EUS-FNA of main duct (MD) (from MD directly or pancreatic cyst/mass arising from MD), and in MT-IPMNs, EUS-FNA of branch duct (BD). The no-intervention group included patients who did not undergo either an ERCP or EUS-FNA. Patients who had EUS without FNA were included in this group. Therefore, the four comparison arms were EUS-FNA alone, ERCP alone, EUS-FNA and ERCP, and no endoscopic intervention. The primary outcomes were surgical adverse events, including 30-day readmission, delayed gastric emptying, pancreatic fistula, abdominal abscess, cardiopulmonary adverse events, and death. Delayed gastric emptying was reclassified as symptoms of nausea, vomiting, bloating, and impaired ability to keep nutrition postoperatively. Pancreatic fistula was diagnosed by elevated amylase in the fluid from peripancreatic drain. Abdominal abscess was diagnosed by imaging studies showing the presence of abscess in the abdomen. Cardiopulmonary adverse events included the presence of systolic blood pressure less than 90, hypoxemia requiring supplemental oxygen, diagnosis of atrial fibrillation with rapid ventricular response, cardiac arrest, pneumonia, acute respiratory distress syndrome, pulmonary embolism, pulmonary edema and/or pleural effusion on imaging studies. The secondary outcomes were immediate post-ERCP adverse events, and immediate post-EUS adverse events.

2.5. Statistical analysis

Multivariable logistic regression analysis was used to compare the postoperative complication risks between intervention and no-intervention groups. Age, BMI, smoking status, ASA class, epithelial subtype, highest grade on histopathology, IPMN size on imaging, type of pancreas surgery, positive margin, large vessel involvement, lymphovascular and perineural invasion, endoscopic procedure, post-ERCP adverse events, and post-EUS adverse events were included as regression variables.

In the analyses, continuous variables were reported as mean with standard deviation (SD). Categorical variables were reported as proportion (%). Odds ratio (OR), P-value and 95 % confidence intervals (CI) were calculated. A P-value of <0.05 was considered to be statistically significant. Bonferroni correction was applied for multiple comparisons. Power analysis was also performed, and the necessary sample size was estimated to be 366 to have 80 % power to detect odds ratio of at least 1.8 for different adverse events, with type I error set at 0.05 [10]. Statistical analyses were performed using R version 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org/).

3. Results

3.1. Baseline characteristics

A total of 369 patients were included in the study. There were 73 patients who had both ERCP and EUS-FNA, 49 patients who had ERCP only, 183 patients who had EUS-FNA only, and 64 patients who had neither ERCP nor EUS-FNA. Among patients who had EUS-FNA, 115 patients had EUS-FNA of MD (40 from MD directly, 75 from the cyst/mass arising from MD) and 141 patients had EUS-FNA of BD. Mean age at the time of surgery was 67.9 years old. The majority of patients included in the study were white (N = 332, 90 %) which represents the demographics of the registries. Patients who had prior pancreatobiliary surgery all had cholecystectomy, and none of them had pancreatic surgery previously. The patient characteristics are summarized in Table 1. There were no significant differences among the four intervention groups.

Table 1.

Demographic and clinical characteristics of patients.^a

Characteristics	Total (N = 369)	ERCP and EUS-FNA (N = 73)	ERCP only (N = 49)	EUS-FNA only (N = 183)	No endoscopic procedures for IPMN (N = 64)	P-value
Age at surgery (years)						0.12
Mean (SD)	67.9 (10.2)	67.1 (10.3)	65.5 (10.3)	69.1 (9.4)	67.2 (11.7)
Min-Max	30–92	30–92	37–85	33–88	43–91
Sex (%)						0.84
Male	181 (49.1 %)	36 (49.3 %)	27 (55.1 %)	88 (48.1 %)	30 (46.9 %)
Female	186 (50.4 %)	37 (50.7 %)	22 (44.9 %)	93 (50.8 %)	34 (53.1 %)
Unknown	2 (0.5 %)	0	0	2 (1.1 %)	0
Ethnicity (%)						0.52
Black	12 (3.3 %)	2 (2.7 %)	2 (4.1 %)	7 (3.8 %)	1 (1.6 %)
White	332 (90.0 %)	69 (94.5 %)	44 (89.8 %)	163 (89.1 %)	56 (87.5 %)
Hispanic	8 (2.2 %)	0	2 (4.1 %)	5 (2.7 %)	1 (1.6 %)
Asian	12 (3.2 %)	2 (2.7 %)	0	5 (2.7 %)	5 (7.8 %)
Unknown	5 (1.4 %)	0	1 (2.0 %)	3 (1.6 %)	1 (1.6 %)
BMI (kg/m2)						0.08
Mean (SD)	26.3 (5.1)	25.3 (4.5)	27.5 (5.6)	26.6 (5.4)	25.6 (4.5)
Min-Max	14.4–48.4	16.8–40.9	16.6–40.3	14.4–48.4	17.2–42.7
Smoking status (%)						0.37
Current	54 (14.6 %)	10 (13.7 %)	12 (24.5 %)	26 (14.2 %)	6 (9.4 %)
Former	142 (38.5 %)	27 (37.0 %)	14 (28.6 %)	73 (39.9 %)	28 (43.8 %)
Never	173 (46.9 %)	36 (49.3 %)	23 (46.9 %)	84 (45.9 %)	30 (46.9 %)
Previous biliary surgery before IPMN diagnosis (%)						0.12
Yes	84 (22.8 %)	24 (32.9 %)	11 (22.4 %)	38 (20.8 %)	11 (17.2 %)
No	285 (77.2 %)	49 (67.1 %)	38 (77.6 %)	145 (79.2 %)	53 (82.8 %)
ASA class (%)						0.25
1	1 (0.3 %)	0	0	0	1 (1.6 %)
2	161 (43.6 %)	39 (53.4 %)	14 (28.6 %)	82 (44.8 %)	26 (40.6 %)
3	141 (38.2 %)	20 (27.4 %)	12 (24.5 %)	83 (45.4 %)	26 (40.6 %)
4	2 (0.5 %)	0	0	2 (1.1 %)	0
Unknown	64 (17.3 %)	14 (19.2 %)	23 (46.9 %)	16 (8.7 %)	11 (17.2 %)

^aPercentages may not sum to 100 because of rounding.

3.2. Factors affecting postoperative outcomes

Postoperative outcomes are summarized in Table 2, including the patient, pathologic, operative, and endoscopic differences among patients who developed adverse events after surgical resection for IPMNs. There were 12.7 % (N = 47) of patients with a 30-day readmission, 13.8 % (N = 51) with delayed gastric emptying, 10.3 % (N = 38) with pancreatic fistula, 5.7 % (N = 21) with abdominal abscess, 11.4 % (N = 42) with cardiopulmonary adverse events, and 1.4 % (N = 5) with death. Age at surgery was a statistically significant factor for developing delayed gastric emptying (P-value = 0.04), and cardiopulmonary adverse events (P-value = 0.02). BMI was statistically significant for developing delayed gastric emptying (P-value = 0.03), abdominal abscess (P-value = 0.03), and cardiopulmonary adverse events (P-value = 0.047). ASA class was statistically significant for developing cardiopulmonary adverse events (P-value <0.001) and death after surgery (P-value <0.001). Epithelial subtype of the IPMN was statistically significant for developing abdominal abscess (P-value = 0.03) and cardiopulmonary adverse events (P-value = 0.01). Specifically, intestinal subtype was associated with a lower risk of abdominal abscess (OR = 0.1, 95 % CI = 0.01–0.61) and cardiopulmonary adverse events (0.64, 95 % CI = 0.27–0.8) than other subtypes. The surgery type for IPMN resection was statistically significant for developing delayed gastric emptying (P-value = 0.004), especially pancreaticoduodenectomy (P-value <0.001 after Bonferroni correction). Lymphovascular invasion on pathology was statistically significant for having 30-day read-mission (P-value = 0.02), and abdominal abscess (P-value = 0.02), whereas having large vessel invasion or perineural invasion was not. EUS-FNA was statistically significant for having cardiopulmonary adverse events (P-value = 0.01), without adjusting for other potential confounding factors. Of note, the time interval between the last endoscopic procedure and surgery was not statistically significant for developing any postoperative adverse events.

Table 2.

Potentially significant variables for postoperative outcomes.^a

Variables	Total (N = 369)	30-day Readmission (N = 47, 12.7 %)	P-value	Delayed Gastric Emptying (N = 51, 13.8 %)	P-value	Pancreatic Fistula (N = 38, 10.3 %)	P-value	Abdominal Abscess (N = 21, 5.7 %)	P-value	Cardio-pulmonary Adverse Events (N = 42, 11.4 %)	P-value	Death (N = 5, 1.4 %)	P-value
Age at surgery (years)			0.53		0.04		0.32		0.73		0.02		0.55
Mean (SD)	67.9 (10.2)	68.7 (9.7)		65.1 (9.9)		66.3 (9.3)		68.6 (7.3)		71.2 (11.3)		70.6 (14.6)
Min-Max	30–92	47–91		44–85		47–83		53–83		37–92		60–88
BMI (kg/m2)			0.81		0.03		0.35		0.03		0.047		0.75
Mean (SD)	26.3 (5.1)	26.1 (4.8)		27.8 (5.7)		27.0 (4.6)		28.6 (6.7)		27.8 (5.7)		25.6 (7.9)
Min-Max	14.4–48.4	14.4–42.7		17.2–42.7		19.3–40.9		21.3–48.4		17.8–42.7		17.8–37.9
Smoking status (%)			0.32		0.66		0.22		0.65		0.06		0.63
Current	54 (14.6 %)	4 (8.5 %)		9 (17.6 %)		2 (5.3 %)		3 (14.3 %)		11 (26.2 %)		0
Former	142 (38.5 %)	17 (36.2 %)		17 (33.3 %)		17 (44.7 %)		10 (47.6 %)		12 (28.6 %)		2 (40 %)
Never	173 (46.9 %)	26 (55.3 %)		25 (49.0 %)		19 (50.0 %)		8 (38.1 %)		19 (45.2 %)		3 (60 %)
ASA class (%)
1	1 (0.3 %)	0	0.47	0	0.70	0	0.92	0	0.97	0	<0.001	0	<0.001
2	161 (43.6 %)	23 (48.9 %)		19 (37.3 %)		17 (44.7 %)		11 (52.4 %)		15 (35.7 %)		2 (40 %)
3	141 (38.2 %)	18 (38.3 %)		22 (43.1 %)		13 (34.2 %)		9 (42.9 %)		16 (38.1 %)		2 (40 %)
4	2 (0.5 %)	1 (2.1 %)		0		0		0		2 (4.8 %)		1 (20 %)
Unknown	64 (17.3 %)	5 (10.6 %)		10 (19.6 %)		8 (21.1 %)		1 (4.8 %)		9 (21.4 %)		0
Epithelial subtype (%)			0.31		0.79		0.11		0.03		0.01		0.77
Gastric	136 (36.9 %)	19 (40 %)		20 (39.2 %)		14 (36.8 %)		13 (61.9 %)		20 (47.6 %)		3 (60 %)
Intestinal	90 (24.4 %)	10 (21.3 %)		15 (29.4 %)		5 (13.2 %)		1 (4.8 %)		9 (21.4 %)		0
Pancreatobiliary	14 (3.8 %)	1 (2.1 %)		1 (2.0 %)		2 (5.3 %)		0		3 (7.1 %)		0
Oncocytic	10 (2.7 %)	0		1 (2.0 %)		1 (2.6 %)		0		0		0
Mixed	47 (12.7 %)	10 (21.3 %)		7 (13.7 %)		10 (26.3 %)		5 (23.8 %)		9 (21.4 %)		1 (20 %)
Unknown	72 (19.5 %)	7 (14.9 %)		7 (13.7 %)		6 (15.8 %)		2 (9.5 %)		1 (2.4 %)		1 (20 %)
Histopathological highest grade (%)			0.08		0.23		0.43		0.90		0.37		0.89
Low grade	147 (39.8 %)	21 (44.7 %)		18 (35.3 %)		19 (50.0 %)		9 (42.9 %)		16 (38.1 %)		2 (40 %)
High grade	103 (27.9 %)	7 (14.9 %)		11 (21.6 %)		9 (23.7 %)		5 (23.8 %)		9 (21.4 %)		1 (20 %)
Invasive	116 (31.4 %)	19 (40.4 %)		21 (41.2 %)		10 (26.3 %)		7 (33.3 %)		17 (40.5 %)		2 (40 %)
Unknown	3 (0.8 %)	0		1 (2.0 %)		0		0		0		0
IPMN size on imaging (mm)			0.93		0.16		0.87		0.29		0.91		0.94
Mean (SD)	31 (20)	31.5 (17)		36.3 (19.2)		31.2 (32.6)		27.1 (12.0)		31.4 (16)		32.4 (13.7)
Min-Max	3–200	5–78		5–80		6–200		10–55		5–78		20–51
Type of pancreatectomy surgery (%)			0.89		0.004		0.50		0.81		0.49		0.19
Pancreaticoduodenectomy	219 (59.3 %)	26 (55.3 %)		41 (80.4 %)		21 (55.3 %)		10 (47.6 %)		23 (54.8 %)		1 (20 %)
Distal	106 (28.7 %)	14 (29.8 %)		4 (7.8 %)		14 (36.8 %)		8 (38.1 %)		15 (35.7 %)		4 (80 %)
Total	14 (3.8 %)	1 (2.1 %)		1 (2.0 %)		0		1 (4.8 %)		2 (4.8 %)		0
Middle segment with Roux-en-Y	14 (3.8 %)	2 (4.3 %)		1 (2.0 %)		2 (5.3 %)		1 (4.8 %)		0		0
Other	4 (1.1 %)	1 (2.1 %)		0		0		0		0		0
Unknown	12 (3.3 %)	3 (6.4 %)		4 (7.8 %)		1 (2.6 %)		1 (4.8 %)		2 (4.8 %)		0
Positive margins b			0.17		0.35		0.28		0.20		0.45		0.54
Yes	25 (6.8 %)	1 (2.1 %)		5 (9.8 %)		1 (2.6 %)		0		4 (9.5 %)		0
No	344 (93.2 %)	46 (97.9 %)		46 (90.2 %)		37 (97.4 %)		21 (100 %)		38 (90.5 %)		5 (100 %)
Large vessel invasion			0.77		0.67		0.46		0.35		0.39		0.13
Yes	19 (5.1 %)	2 (4.3 %)		2 (3.9 %)		1 (2.6 %)		2 (9.5 %)		1 (2.4 %)		1 (20 %)
No	350 (94.9 %)	45 (95.7 %)		49 (96.1 %)		37 (97.4 %)		19 (90.5 %)		41 (97.6 %)		4 (80 %)
			0.02		0.13		0.70		0.02		0.92		0.24
Lymphovascular invasion
Yes	25 (6.8 %)	7 (14.9 %)		6 (11.8 %)		2 (5.3 %)		4 (19.0 %)		3 (7.1 %)		1 (20 %)
No	344 (93.2 %)	40 (85.1 %)		45 (88.2 %)		36 (94.7 %)		17 (81.0 %)		39 (92.9 %)		4 (80 %)
Perineural invasion			0.96		0.28		0.79		0.22		0.95		0.73
Yes	54 (14.6 %)	7 (14.9 %)		10 (19.6 %)		5 (13.2 %)		5 (23.8 %)		6 (14.3 %)		1 (20 %)
No	315 (85.4 %)	40 (85.1 %)		41 (80.4 %)		33 (86.8 %)		16 (76.2 %)		36 (85.7 %)		4 (80 %)
ERCP			0.31		0.49		0.84		0.06		0.83		0.74
Yes	122 (33.1 %)	12 (25.5 %)		19 (37.3 %)		12 (31.6 %)		3 (14.3 %)		15 (35.7 %)		2 (40 %)
No	247 (66.9 %)	35 (74.5 %)		32 (62.7 %)		26 (68.4 %)		18 (85.7 %)		27 (64.3 %)		3 (60 %)
Post-ERCP adverse events (%)			0.052		0.07		<0.001		0.93		0.82		0.99
Pancreatitis	5 (1.4 %)	0		0		0		0		0		0
Hemorrhage	1 (0.3 %)	1 (2.1 %)		0		1 (2.6 %)		0		0		0
Bile leak	1 (0.3 %)	0		1 (2.0 %)		1 (2.6 %)		0		0		0
None	362 (98.1 %)	46 (97.9 %)		50 (98.0 %)		36 (94.7 %)		21 (100 %)		42 (100 %)		5 (100 %)
EUS-FNA (%)			0.23		0.90		0.88		0.22		0.01		0.71
Yes, MD aspirated	121 (32.8 %)	19 (40.4 %)		17 (33.3 %)		11 (28.9 %)		3 (14.3 %)		18 (42.9 %)		2 (40 %)
Yes, BD aspirated	139 (37.7 %)	19 (40.4 %)		19 (37.3 %)		14 (36.8 %)		11 (52.4 %)		7 (16.7 %)		1 (20 %)
No	109 (29.5 %)	9 (19.1 %)		15 (29.4 %)		13 (34.2 %)		7 (33.3 %)		17 (40.5 %)		2 (40 %)
Post-EUS adverse events			0.46		0.51		0.50		0.62		0.47		0.81
Yes	4 (1.1 %)	1 (2.1 %)		1 (2.0 %)		0		0		0		0
No	365 (98.9 %)	46 (97.9 %)		50 (98 %)		38 (100 %)		21 (100 %)		42 (100 %)		5 (100 %)
Interval between endoscopic procedure and surgery (days)			0.68		0.65		0.68		0.83		0.73		0.95
Mean (SD)	275.8 (2508.8)	111 (200)		78.7 (107)		88.2 (127)		140 (288)		107 (157)		187 (214)
Min-Max	2–41381	4–1134		11–587		8–653		14–1134		2–627		60–434

^aPercentages may not sum to 100 because of rounding.

^bPositive margins on pathology include positive pancreatic margins, biliary margins, retroperitoneal margins, or mesenteric margins for IPMNs and invasive carcinoma.

There were 7 patients with immediate post-ERCP adverse events (5 (1.4 %) with pancreatitis, 1 (0.3 %) with hemorrhage, 1 (0.3 %) with bile leak), and 4 patients (1.1 %) had immediate post-EUS adverse events (pancreatitis). Having immediate post-ERCP adverse events was statistically significant for developing pancreatic fistula after surgery (P-value <0.001), especially for hemorrhage or bile leak (P-value = 0.02 for both after Bonferroni correction). Having immediate post-EUS adverse events was not a statistically significant factor for developing other postoperative adverse events.

3.3. Postoperative adverse events by intervention groups

Postoperative adverse events after ERCP and EUS-FNA were evaluated independently. After adjusting for confounding factors, there were no significant differences between the ERCP and no-ERCP groups for 30-day readmission (odds ratio (OR) = 0.8, 95 % confidence interval (CI) = 0.3–2.2, P-value = 0.7), delayed gastric emptying (OR = 1.4, 95 % CI = 0.4–4.4, P-value = 0.6), pancreatic fistula (OR = 1_, 95 % CI = 0.3_–3.1, P-value = 0.9), abdominal abscess (OR = 0.5, 95 % CI = 0.1–2.7, P-value = 0.5), cardiopulmonary adverse events (OR = 1.9, 95 % CI = 0.5–6.8, P-value = 0.3), or death (OR = 10, 95 % CI = 0.01–20, P-value = 0.9), when using no-ERCP group as the reference group (Table 3).

Table 3.

Odds ratio (OR) of post-surgical adverse events in patients who received and did not receive ERCP, and in patients who received and did not receive EUS-FNA, respectively^a.

	30-day Readmission			Delayed Gastric Emptying			Pancreatic Fistula			Abdominal Abscess			Cardiopulmonary Adverse Events			Death
	N 47	OR (95 % CI)	P-value	N 51	OR (95 % CI)	P-value	N 38	OR (95 % CI)	P-value	N 21	OR (95 % CI)	P-value	N 42	OR (95 % CI)	P-value	N 5	OR (95 % CI)	P-value
ERCP
Yes	12	0.8 (0.3–2.2)	0.7	19	1.4 (0.4–4.4)	0.6	12	1 (0.3–3.1)	0.9	3	0.5 (0.1–2.7)	0.5	15	1.9 (0.5–6.8)	0.3	2	10 (0.01–20)	0.9
No	35		–	32	–	–	26	–	–	18	–	–	27	–	–	3	–	–
EUS-FNA
Yes, MD aspirated from duct	4	2 (0.4–9.1)	0.4	3	0.41 (0.02–3.7)	0.5	3	0.4 (0.02–2.8)	0.4	1	0.2 (0.01–2.3)	0.3	3	0.1 (0.01–1.1)	0.07	0	–	–
Yes, MD aspirated from cyst/mass	12	1.6 (0.5–5.7)	0.5	14	3 (0.8–13.2)	0.1	8	1.1 (0.3–3.9)	0.8	2	0.3 (0.02–1.8)	0.2	14	0.7 (0.2–2.8)	0.6	2	8 (0.01 – 15)	0.9
Yes, BD aspirated	22	1.4 (0.5–4.4)	0.5	19	1.5 (0.4–5.7)	0.5	14	0.7 (0.2–1.9)	0.4	11	1.1 (0.3–4.9)	0.9	8	0.3 (0.1–0.9)	0.04	1	0.03 (0.01–25)	0.9
No	9	–	–	15	–	–	13	–	–	7	–	–	17	–	–	2	–	–

^aAdjusted for age, BMI, smoking status, ASA class, epithelial subtype, histopathological grade, IPMN size, type of pancreatectomy surgery, positive margins, vascular/ lymphovascular/perineural invasions.

In addition, no significant differences were observed for the EUS-FNA and no-EUS-FNA groups, except for cardiopulmonary adverse events where EUS-FNA with BD aspiration was weakly associated with a lower risk (OR = 0.3, 95 % CI = 0.1–0.9, P-value = 0.04), using no-EUS-FNA group as the reference group. Specifically, EUS-FNA with MD aspiration from the main pancreatic duct directly or cyst/mass arising from MD showed no significant increase in any specific adverse events (Table 3).

4. Discussion

EUS-FNA and ERCP are often used in the diagnosis of IPMN before surgery to further classify risk features and treat biliary or pancreatic duct obstruction. However, the impact of preoperative endoscopic procedures on surgical outcomes remains to be investigated, especially procedures that directly manipulate the main pancreatic duct. In the present study, we evaluated the surgical outcomes by comparing postoperative adverse event rates in patients who underwent EUS-FNA and/or ERCP and those who did not, before they received surgical resection for MD-IPMNs or MT-IPMNs. There was no significant difference in surgical outcomes between patients who had ERCP and those who did not, and between patients who had EUS-FNA and those who did not. Particularly, patients who had EUS-FNA with MD aspiration either from the main duct directly or cyst/mass arising from MD did not have increased rates of any postoperative adverse events.

Results of the present study are consistent with the existing literature [11], demonstrating that preoperative EUS-FNA and ERCP were not associated with worse surgical outcomes. Furthermore, our study analyzed specific postsurgical adverse events, including 30-day readmission, delayed gastric emptying, pancreatic fistula, abdominal abscess, cardiopulmonary adverse events, and death, which were some of the most common and important adverse events when assessing surgical outcomes. We also analyzed immediate post-ERCP and post-EUS adverse events, including pancreatitis, bile leak and hemorrhage. Their risks were low, and all were <1.5 %. Interestingly, in patients with immediate post-ERCP adverse events, the risk of postsurgical pancreatic fistula was increased. While the reason why post-ERCP adverse events increased the risk of pancreatic fistulas is not clear, we postulate that this may be due to ERCPs leading to colonization of the pancreaticobiliary tree, thus causing an increased risk of infected fistulas, which may be clinically apparent. Another possibility is that adverse events such as a biliary leak may lead to scarring or adhesions that increased the technical challenge of the surgery. While the cause is not clear, it is important to note that immediate post-ERCP adverse events may increase the risk of postsurgical pancreatic fistula and any precautions to minimize them should be taken.

Factors that were found to be statistically significant for developing postoperative adverse events include age, BMI, ASA class, epithelial subtype, surgery type, and lymphovascular invasion on pathology. While advanced age, BMI >25, and increasing ASA class are known risk factors for worse postoperative outcomes after pancreatic surgery [12–14], our results also showed that intestinal subtype was associated with a lower risk of abdominal abscess and cardiopulmonary adverse events than other subtypes. Previous research suggested that epithelial subtypes play a role in the biology and prognosis of invasive IPMNs [15]. Colloid carcinomas often arise from intestinal-type IPMNs and are associated with markedly better outcome than tubular carcinomas, which primarily originate from gastric-type IPMNs. In addition, gastric-type IPMNs usually involve the BD, and the overall survival of patients with adenocarcinoma from BD-IPMNs was shown to be shorter than those with MD-IPMNs, which are correlated with intestinal- or pancreatobiliary-type IPMNs [15]. Specific reasons why epithelial subtypes are associated with certain postoperative adverse events but not others warrant further study. Pancreaticoduodenectomy was associated with a higher risk of delayed gastric emptying, consistent with prior studies that delayed gastric emptying is frequently observed after pancreaticoduodenectomy [16,17]. There is no clear mechanism that might explain why lymphovascular invasion, not large vessel or perineural invasion on pathology, would be associated with higher risks of postoperative adverse events, and further research is needed.

While not directly examined in our study, one of the common concerns regarding the safety of EUS-FNA for IPMNs is needle tract seeding, because FNA could cause leakage of the cystic content. However, many studies have shown that EUS-FNA before surgery is not associated with a higher risk of peritoneal seeding in patients who had IPMN or pancreatic cancer [18–20]. In addition, a previous study by Suzuki et al. showed that EUS-FNA did not have an impact on the relapse-free survival, overall survival or recurrence patent in patients who had IPMN-derived adenocarcinoma [21]. Similarly, Maruta et al. showed that there were no significant differences in surgical curability, peritoneal dissemination, median survival times between the FNA and non-FNA groups in patients who underwent surgical resection for pancreatic cancer [22]. These results further support the safety of EUS-FNA in the evaluation of IPMNs.

Indeed, EUS-FNA has many advantages in the diagnosis and risk stratification of IPMNs. EUS-FNA has aided in the diagnosis of mucinous cysts by utilizing CEA level of >192 to differentiate mucinous from nonmucinous pancreatic cysts [23]. Furthermore, it has a 77 % sensitivity in detecting pancreatic adenocarcinoma, and a 93 % sensitivity when atypical and suspicious cytology is reclassified as diagnostic of malignancy [24]. With the advent of next-generation sequencing, EUS-FNA allows the collection of cystic fluid to find genetic mutations that are highly sensitive and specific for different pancreatic cyst types as well as advanced neoplasia [25]. As shown in our study, sampling the main duct with EUS-FNA was safe and did not affect postoperative outcomes. Therefore, EUS-FNA might be a better option than through-the-needle biopsy (TTNB) to evaluate pancreatic cysts when the suspicion for IPMN is high, as IPMN diagnosis is associated with an increased risk of adverse events after TTNB for pancreatic cystic lesions [26]. Additional diagnostic options such as contrast-enhanced EUS can also aid in the characterization of mural nodules, a risk factor for malignancy, and potentially eliminate the need for FNA [27].

The strengths of this study include assessing the effect of EUS-FNA with MD aspiration, including direct aspiration from MD or aspiration from cyst/mass arising from MD, which was not directly studied in prior literature. In addition, we used a bi-center cohort enrolled in tertiary care centers, spanning 19 years. This study also has some limitations. Although data were collected prospectively, the retrospective study design in analysis could lead to biases in patient selection and treatment strategies. Moreover, only patients who eventually underwent surgical resection for IPMNs were included in the study. Future studies should also examine the adverse event rates in patients who underwent EUS-FNA and/or ERCP for IPMNs but did not eventually have surgery.

5. Conclusions

This bi-center retrospective study revealed no association between preoperative ERCP or EUS-FNA and adverse events in patients undergoing surgical resection for MD-IPMNs or mixed-type IPMNs. Particularly, EUS-FNA with direct aspiration of the main duct or the cyst/mass arising from the main duct was not associated with an increased risk of adverse surgical events. ERCP and EUS-FNA can be safely used in the evaluation of IPMNs in selected cases.