Long-Term Surveillance of Branch-Duct Intraductal Papillary Mucinous Neoplasms without Worrisome or High-Risk Features

Abstract

Introduction:

Long-term data evaluating clinical outcomes in patients with branch-duct IPMN (BD-IPMN) without high-risk stigmata (HRS) or worrisome features (WF) remain limited.

Methods:

This observational cohort study included all patients diagnosed with BD-IPMN without HRS or WF between 2003 and 2019 who were enrolled in a prospective surveillance program. Time to progression analysis was performed using a cumulative incidence function plot and survival analysis was conducted using Kaplan-Meier.

Results:

Median follow-up time for the 267 patient cohort was 44.5 months (IQR:24.1–72.2). Radiographic cyst growth was observed in 123 (46.1%) patients; 65 (24.3%) patients progressed to WF/HRS. Twenty-six (9.7%) patients were selected for resection during surveillance: 21 (80.8%) WF, 4 (15.4%) HRS; 1 (3.9%) transformed to mixed-duct. Of all the patients who underwent resection, 5 (19.2%) had adenocarcinoma, and 1 (3.8%) had carcinoma-in-situ. The probability of any radiographic progression was 21.3% (5-year) and 51.3% (10-year). For the entire cohort, there was 1.1% mortality secondary to pancreatic adenocarcinoma and 8.2% all-cause mortality. The 5-year overall survival rate was 91.5%, and at 10 years, 81.5%.

Conclusion:

Approximately one in four patients with non-worrisome BD-IPMN have progression to WF/HRS stigmata during surveillance. However, the risk of malignant transformation remains low. Surveillance strategy remains prudent in this patient population.

INTRODUCTION

Pancreatic cystic neoplasms have become an increasingly common incidental finding secondary to advances in modalities and frequency of cross-sectional imaging.^1–6 Intraductal papillary mucinous neoplasms (IPMN) account for the vast majority of incidentally found pancreatic cystic lesions.^2,3,6 There are three subtypes of IPMNs, main-duct (MD-IPMN), branch-duct (BD-IPMN), and mixed-type (MT-IPMN),⁶ which carry variable risk for malignancy or malignant transformation.^3,7 Although BD-IPMNs have the lowest risk of malignant transformation (3–46%),^2,6,8 they are the most common IPMN type observed and their management has been persistently debated.^7,8

Guidelines for IPMN management are based on cyst characteristics associated with concern for transformation to malignancy including cyst size, presence of a solid component, and main pancreatic duct dilation. The updated International Consensus Guidelines are commonly used by pancreatic surgeons in the management of BD-IPMNs.⁹ These guidelines define worrisome features (WF) and high-risk stigmata (HRS) to help guide decision-making regarding recommended treatment. According to these guidelines, BD-IPMNs that do not present with WF or HRS are stratified into surveillance categories based on cyst size.⁹ While these recommendations are based on the best evidence available, long-term data supporting management guidelines for patients with branch-duct IPMN (BD-IPMN) without high-risk stigmata or worrisome features remain limited and debated.^6,8

Criticism of management strategies for BD-IPMN focus on the frequency and duration of surveillance and the utility of resection, with ongoing debate regarding treatment strategies given differences in long-term outcome data.^2,5,10 Indefinite surveillance can result in years of imaging without reassurance of safety from progression.² Previous investigations have reported a risk of progression at 5 years ranging from 10%−70% with higher rates of progression at greater than 5 years.^1,11 Surgical management may eliminate risk of progression, however, pancreatic resection remains a major operation with significant short-term morbidity and the potential for long-term pancreatic endocrine and exocrine insufficiency affecting patients’ quality of life.^2,8,12 Given the lack of long-term granular surveillance data, we aimed to determine the long-term risk of progression during surveillance in patients with BD-IPMN without worrisome features or high-risk stigmata.

METHODS

Study Design and Population

This observational cohort study included all patients 18 years of age and older who were evaluated for BD-IPMN in the University of Virginia pancreatic surgery clinic between 2003 and 2019. All patients were enrolled in a prospective surveillance program and data were maintained in an institutional database. Only patients with a radiographic diagnosis of BD-IPMN without worrisome features or high-risk stigmata were included in the study. Patients who had high-risk stigmata, worrisome features, and a mixed or main duct component at the time of initial evaluation or within three months of diagnosis of BD-IPMN were excluded. Patients who underwent surgical resection within 6 months of diagnosis were also excluded from the study. Our group’s approach to patients with pancreatic cystic neoplasms has been traditionally conservative, offering resection to patients in whom the risk of malignancy was considered to outweigh the risks of resection. Prior to the international consensus guidelines published in 2006¹³, recognized lower risk of malignancy in patients with BD-IPMN less than 3 cm and without mural nodules led to the recommendation for surveillance in these patients.^14–16 At present, our group uses the updated 2017 International Consensus Guidelines⁹ to help guide decision-making in patients with IPMN. The University of Virginia Institutional Review Board for Health Sciences Research reviewed this study, and it was considered exempt (IRB-HSR# 21609).

Study Objectives and Variables

The primary objective of this study was to determine the risk of BD-IPMN progression in patients presenting without worrisome features or high-risk stigmata at the time of diagnosis. Study outcomes included cyst growth (≥ 2 mm from baseline), progression to WF or HRS and progression to malignancy (carcinoma-in-situ and invasive adenocarcinoma). Worrisome features were defined based on both clinical and radiologic factors: the presence of pancreatitis, elevated CA 19–9, cyst ≥ 3 cm, enhancing mural < 5 mm, thickened/enhanced cyst walls, main duct size 5–9 mm, changes in duct caliber, or lymphadenopathy. High-risk stigmata were defined as enhancing solid component ≥ 5 mm, main duct dilation ≥ 10 mm, or obstructive jaundice.⁹

Follow-up time was defined as the interval between initial diagnosis and most recent surveillance imaging or selection for surgical resection. Overall mortality was stratified as all-cause mortality and disease-specific mortality from pancreatic cancer. Demographic and comorbid condition variables included in the study were age, sex, race (White, Black, Asian, and unknown), body mass index, diabetes, coronary artery disease, chronic obstructive pulmonary disease, and chronic kidney disease. Perioperative variables included type pancreatic resection, postoperative complications (including pancreatic fistula, bile leak, surgical site infections), and 30-day and 90-day mortality.

Statistical Analysis

Data were presented as median (interquartile range) for continuous variables and frequency (percentage) for categorical variables. A time-to-progression analysis was performed using a cumulative incidence function plot accounting for mortality as a competing risk. An overall survival analysis was conducted using a Kaplan-Meier plot. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

A total of 267 patients with BD-IPMN were included in the study. Patient demographics and preoperative comorbid conditions are included in Table 1. The majority of patients were White (90.0%) and female (57.3%) with a median age of 66.0 (IQR: 58.0–73.0) years old at the time of diagnosis. The median BMI was 26.6 (IQR: 23.7–31.8). The most common comorbid condition was diabetes mellitus (24.0%).

Table 1.

Characteristics among patients with branch duct intraductal papillary mucinous neoplasm (BD-IPMN) without worrisome features or high-risk stigmata selected for surveillance.

	Surveillance cohort (N=267)
Age at diagnosis, years, median (IQR)	66.0 (58.0–73.0)
Male sex	114 (42.7)
Race
White	243 (90.0)
Black	13 (4.9)
Asian	1 (0.4)
Unknown	10 (3.8)
Body mass index, kg/m2, median (IQR)	26.6 (23.7–31.8)
Diabetes	64 (24.0)
CAD	26 (9.7)
COPD	16 (6.0)
CKD	30 (11.2)
Alcohol use	128 (47.9)
Tobacco use	117 (43.8)

Data presented as N (%) unless otherwise specified.

Median cyst size at the beginning of the surveillance period was 15 mm (IQR: 9–22). The median follow-up time for this surveillance cohort was 44.5 months (IQR: 24.1–72.2). Clinical follow-up data are summarized in Table 2. Median cyst size during the last available surveillance period was 15 mm (IQR: 10–27) for the entire cohort and 29 mm (IQR: 20–42) for patients selected for resection. Of the entire cohort, 111 (41.6%) patients underwent EUS during the surveillance period with 54.1% of the EUS patients undergoing FNA/biopsy at the time of the procedure. The EUS findings included: 43 (71.7%) normal, 13 (21.7%) atypia, 2 (3.3%) dysplasia, 2 (3.3%) adenocarcinoma.

Table 2.

Clinical outcomes among patients with branch duct intraductal papillary mucinous neoplasm (BD-IPMN) selected for surveillance.

Entire surveillance cohort	267 (100)
Cyst size at the beginning of surveillance, mm, median (IQR)	15 (9–22)
Last known cyst size, mm, median (IQR)	17 (10–27)
Follow-up time, months, median (IQR)	44.5 (24.1–72.2)
EUS	111 (41.6)
FNA/Biopsy	60 (54.1)
Normal	43 (71.7)
Atypia	13 (21.7)
Dysplasia	2 (3.3)
Adenocarcinoma	2 (3.3)
Growth	123 (46.1)
cyst growth without progression	69 (56.1)
Progression *	65 (24.3)
Malignancy	6 (2.2)
Death from any cause	22 (8.2)
Death from pancreatic adenocarcinoma	3 (1.1)
Patients selected for surgical resection	26 (9.7)
Radiographic/clinical criteria of those selected for surgical resection
Cyst size, median, IQR	29 (20–42)
Worrisome features	21 (80.8)
Cyst > 3cm	12 (46.2)
Pancreatitis	8 (30.8)
Main duct 5–9mm	8 (30.8)
Thickened enhanced cyst walls	5 (19.2)
Growth > 5mm over 2 years	7 (26.9)
Mural nodule < 5mm	4 (15.4)
Increased CA 19–9	2 (7.7)
Lymphadenopathy	0 (0)
High risk stigmata	4 (15.4)
Mural nodule > 5mm	2 (7.7)
Main duct > 10mm	1 (3.9)
Jaundice	1 (3.9)
Mixed duct	1 (3.9)
EUS	26 (100)
FNA/Biopsy	18 (69.2)
Normal	9 (50.0)
Atypia	6 (33.3)
Dysplasia	1 (5.6)
Adenocarcinoma	2 (11.1)

Data presented as N (%) unless otherwise specified.

EUS = endoscopic ultrasonography

FNA = fine needle aspiration

^*Progression is defined as advancement to worrisome features (cyst > 3 cm, pancreatitis, main duct 5–9 mm, thickened cyst walls, growth > 5 mm over two years, mural nodule < 5 mm, increase in CA 19–9 levels, lymphadenopathy), high-risk stigmata (mural nodule > 5 mm, main duct > 10 mm, presence of jaundice), or mixed duct.

During the surveillance period, radiographic growth of lesions≥ 2 mm was observed in 123 (46.1%) patients. Of these 123 patients with growth ≥ 2mm, despite having interval growth, 56.1% did not have concurrent features of progression. A total of 65 (24.3%) patients had radiographic progression to WF or HRS, and 6 (2.2%) progressed to pancreatic cancer or carcinoma in situ. The probability of progression for the entire cohort was 5% at one year, 21.4% at 5 years, and 50.4% at 10 years (Figure 1).

Figure 1.

Time to progression among patients selected for surveillance of branch duct intraductal papillary mucinous neoplasm (BD-IPMN).

From the entire surveillance cohort, 26 (9.7%) patients were selected for surgical resection based on imaging findings consistent with progression (Table 2). Of the 26 patients, 100% had undergone an EUS with 18 (69.2%) having an FNA/biopsy. The EUS results of those selected for surgery included: 9 (50.0%) normal, 6 (33.3%) atypia, 1 (5.6%) dysplasia, and 2 (11.1%) adenocarcinoma. Of these 26 patients, 21 (80.8%) developed worrisome features, 4 (15.4%) developed high-risk stigmata, and 1 (3.9%) progressed to mixed duct IPMN. The most common worrisome features observed were cyst> 3 cm (46.2%), pancreatitis (30.8%), main duct 5–9 mm (26.9%), and growth > 5 mm over 2 years (26.9%). Of the 26 patients who had resection, the majority of patients (57.5%) had pancreatoduodenectomy, and 90-day mortality was zero. Data summarizing the type of resection, postoperative complications, and final pathology are enumerated in Table 3. From this 26-patient resection cohort, 18 patients had IPMN with low-grade and moderate-grade dysplasia, 1 patient had carcinoma in situ within IPMN, and 5 of 27 patients had pancreatic adenocarcinoma. Other pathologic diagnoses included 1 acinar cell cystadenoma and 1 chronic pancreatitis with intraparenchymal cystic component.

Table 3.

Operative management, complications, and pathologic findings

Patients selected for surgical resection	26 (100)
Type of Resection
Pancreatoduodenectomy	15 (57.5)
Distal pancreatectomy	10 (38.5)
Total pancreatectomy	1 (4)
Surgical Pathologic findings
IPMN	18 (69.2)
Carcinoma in situ	1 (3.8)
PDAC	5 (19.2)
Acinar cell cystadenoma	1 (3.8)
Chronic pancreatitis	1 (3.8)
Complications
Pancreatic fistula	3 (11.5)
Bile leak	0 (0.0)
Delayed gastric emptying	2 (7.7)
Superficial surgical site infection	1 (3.8)
Deep/organ space surgical site infection	3 (11.5)
30-day mortality	0 (0.0)
90-day mortality	0 (0.0)

Data presented as N (%)

IPMN = intraductal papillary mucinous neoplasm

PDAC = pancreatic ductal adenocarcinoma

Among patients selected for surveillance, the probability of developing HRS remained very low (1.3%) for the first 5 years (Figure 2). The probability of developing HRS at 10 years was 2.2%. Among patients selected for surveillance, the probability of developing worrisome features was 5.0% at one year, 21.1% at 5 years, and 51.0% at 10 years (Figure 3).

Figure 2.

Time to high-risk stigmata among patients selected for surveillance.

Figure 3.

Time to worrisome features among patients selected for surveillance.

The all-cause mortality rate was 8.2% for the entire surveillance cohort and death from pancreatic adenocarcinoma occurred in 1.1% (Table 2). Overall survival was 100% at one year, 91.5% at 5 years, and 81.4% at 10 years (Figure 4).

Figure 4.

Overall survival among patients selected for surveillance of brand duct intraductal papillary mucinous neoplasm (BD-IPMN).

DISCUSSION

Approximately one in four patients diagnosed with BD-IPMN without worrisome features or high-risk stigmata have radiographic progression to worrisome features or high-risk stigmata as defined by consensus guidelines during the long-term follow-up period. However, very few patients develop pancreatic adenocarcinoma. Although almost half of the patients had cyst growth of 2 mm or greater for up to 10 years of surveillance, only 6 patients required resection for pancreatic cancer or carcinoma in situ, and only 2 patients died from pancreatic cancer. The very low progression rate to high-risk stigmata or carcinoma during long-term surveillance confirms current practice guidelines for patients with BD-IPMN without worrisome features or high-risk stigmata.

While radiographic cyst growth was common in the study population, progression to worrisome features or high-risk stigmata was less common, and selection for surgical resection was even rarer. Data reporting proportional progression among patients with low-risk IPMN are variable, contributing to the debate of whether or not to continue surveillance and, if so, for how long.^6,9 Our findings of progression are similar to published studies^17–21, with frequent radiographic progression but a low risk of progression to high-risk IPMN and/or adenocarcinoma. Two recent cohorts have reported a high proportional risk of radiographic progression from BD-IPMN without worrisome or HRS: 3.7%, 23.4%, and 43.3% at 1, 5, and 10 years, respectively¹ and even higher cumulative risk of radiographic progression 70% at 4 years and 97.5% at 10 years.¹¹ Given the long-term risk of radiographic progression in our study and these published studies, long-term surveillance is warranted.

A rapid growth rate of BD-IPMNs has been suggested as an independent factor associated with the development of pancreatic cancer.^3,20,22 Different growth rates have been described. For example, in a cohort of over one thousand patients with BD-IPMN without worrisome features or HRS, a growth rate of > 2.5 mm per year was associated with developing pancreatic cancer²², while another study reported that the risk of developing any worrisome features increased with every 1 mm increase in cyst size.⁴ Although it is essential to consider growth in surveillance, many BD-IPMNs do not grow during reported follow up as is evident from our data and published studies.⁴ However, there is significant controversy about ongoing surveillance among patients with BD-IPMN without worrisome features or high-risk stigmata or any progression during the initial 5 years of surveillance with AGA guidelines recommending cessation of surveillance.^3,23,24 This recommendation is controversial, given the lack of long-term data demonstrating its efficacy and known chance of progression and cyst growth beyond the 5- and 10-year surveillance periods.^1,11,25 Our data similarly demonstrate persistent radiographic progression to worrisome features and high-risk stigmata beyond the initial 5-year surveillance period.

The low rate of malignant transformation and relatively slow progression of BD-IPMNs noted in our data and published literature^3,17,19,26 have been used to question the frequency of surveillance. Short follow-up periods, frequently less than 5 years, have been used to question recommendations from published data.³ The median follow-up time in our study is within the typical surveillance period of published work, with a maximum follow-up of 17 years, with a reported risk of malignant transformation within the reported range of 0 to 5.5%.^{1,4,5,8,21,26,27} Few studies with greater than 10 year follow-up period report divergent results. Of an 804 patient cohort followed for greater than 5 years in Tokyo, 12% developed pancreatic cancer during 15-year follow-up from index BD-IPMN diagnosis.⁸ However, in another recent multi-institutional study from California, only 1% of patients with BD-IPMN developed pancreatic cancer during 12 year surveillance period.¹⁹ With such a significant discrepancy between published studies in proportional risk of malignant transformation during long-term surveillance, ongoing surveillance in this patient population appears prudent.

A total of 9.7% of the patient cohort in our study had progression and were selected for resection. Of those patients who had resection, most cystic neoplasms had one or more radiographic worrisome features; however, the majority did not have pathologic diagnosis of carcinoma or carcinoma in situ. These findings are in congruency with other recent reports demonstrating the frequency of resection based on progression to radiographic worrisome features with an overall low prevalence of malignancy during pathologic review.²⁰ Other recent studies have reported the risk of adenocarcinoma or carcinoma in situ as high as 20% among patients with BD-IPMN without preoperative main duct component or high-risk cytology.¹⁰ One of the criticisms of the current indefinite surveillance guidelines is that surveillance may lead to the overtreatment of IPMNs and the resection of benign pathology.² While risks and benefits of resection need to be individualized to each patient, arguably, the goal of resection should be prevention of pancreatic cancer in high-risk patients, not early diagnosis of pancreatic cancer in patients who could have had resection prior to developing malignancy in the first place. Risk stratification of patients based on radiographic findings alone remains challenging, and evolving molecular analyses are likely to help with risk stratification in the future.²⁸ While we have not used fluid genomic analysis for patients in this study, results of recent Next Generation Sequencing (NGS) studies make NGS cyst fluid analysis an attractive adjunct to current clinical decision making.^29,30

Balancing the operative risk of pancreatic resection and the associated long-term increased chance of endocrine/exocrine insufficiency with the risk of BD-IPMN progression to malignancy and the associated chance of cancer-related mortality remains a challenge. Mortality from progression to pancreatic cancer in patients with BD-IPMNs without worrisome features or high-risk stigmata remains low. The 98.9% cancer-specific survival in our study is within the 97–100% proportional survival in recent studies with surveillance for 5 years or longer.^{1,11,18,19,31,32} Similarly, more patients die during surveillance from other causes than pancreatic cancer.^11,31,32

This study is limited by a retrospective cohort, single-institution design. However, patients were enrolled in a prospective surveillance program and the granularity of data collection and individual review of each case strengthened data reliability and outcome measures. The study design, which excluded patients who were selected for resection within 6 months of index diagnosis, avoids including patients who were deemed high-risk at presentation or potentially had discordant results with studies performed at referring institutions and not available for review during data collection.

CONCLUSION

Approximately one in two patients with BD-IPMN without worrisome features or high-risk stigmata have radiographic progression during the surveillance period, and one in four patients develop worrisome features or high-risk stigmata. While the risk of adenocarcinoma remains low, long-term surveillance in this patient population is warranted. Further investigations should focus on incorporating molecular analyses into risk-stratification strategy among patients with BD-IPMN.

Synopsis:

This study evaluated risk of progression among patients with BD-IPMN without worrisome features or high-risk stigmata. Approximately one in two patients with BD-IPMN without worrisome features or high-risk stigmata developed radiographic progression during surveillance period and one in four patients developed worrisome features or high-risk stigmata. While the risk of adenocarcinoma remains low, long-term surveillance in this patient population is warranted.