Management of Intraductal Papillary Mucinous Neoplasms: Controversies in Guidelines and Future Perspectives

Purpose of review Management of intraductal papillary mucinous neoplasm (IPMN) is currently based on consensus, in the absence of evidence-based guidelines. In recent years, several consensus guidelines have been published, with distinct management strategies. In this review, we will discuss these discrepancies, in order to guide treating physicians in clinical management. Recent findings The detection rate of pancreatic cysts has increased substantially with the expanded use of high-quality imaging techniques to up to 45%. Of these cysts, 24–82% are IPMNs, which harbour a malignant potential. Timely detection of high-risk lesions is therefore of great importance. Surgical management is based on the presence of clinical and morphological high-risk features, yet the majority of resected specimens appear to be low risk. Summary International collaboration and incentive large-scale prospective registries of individuals undergoing cyst surveillance are needed to accumulate unbiased data and develop evidence-based guidelines. Additionally, development of non-invasive, accurate diagnostic tools (e.g. biomarkers) is needed to differentiate between neoplastic and non-neoplastic pancreatic cysts and detect malignant transformation at an early stage (i.e. high-grade dysplasia).


Introduction
Intraductal papillary mucinous neoplasm (IPMN) is a pancreatic cystic lesion originating from intraductal growth of mucin producing cells. In 1980, Ohhashi et al. [1] were the first to describe IPMN. In 1996, it was recognised as a separate entity [2,3]. The increased detection and awareness of IPMNs led to the development of several, mainly consensus-based, periodically revised national and international guidelines [4•, 5•, 6•, 7•, 8•, 9•, 10•]. Notably, evidence is mainly based on surgical cohorts and information on patients managed conservatively is limited.

Classification
Based on localization and extent, three subtypes can be identified; main-duct (MD-IPMN), branch-duct (BD-IPMN) and mixed-type IPMN (MT-IPMN). Every subtype exhibits a certain risk of malignancy and requires a specific therapeutic approach.
IPMN is also classified according to its cellular morphology as gastric, intestinal, cholangio-papillary or oncocytic type. This classification is based on mucin (MUC) gene expression, architecture and cytology, yet different subtypes can be seen in the same cyst. Each type exhibits a particular risk of malignancy (Table 1).

Risk factors
Both the risk of IPMN development and malignant degeneration increase with age [12,15,17,19,20,31]. The mean age at time of IPMN detection is 65 years.
There is a small male gender predisposition [12,19,20]. Also, lifestyle is of influence, as smoking and alcohol abuse increase the risk of having high-risk and worrisome features [11,31]. Increased BMI and the associated presence of abdominal fat are known to play a role in the development of other pancreatic diseases (e.g. type-2 diabetes mellitus (DM) and pancreatic ductal adenocarcinoma (PDAC)), due to fatty infiltration and inflammation [32,33]. Yet, knowledge about the relation between abdominal fat, IPMN and subsequent malignant transformation is limited. Sturm et al. (2013) [34] found a relation between severe obesity (BMI ≥ 35) and an increased risk of malignant transformation in IPMN (OR 10.1, 95% CI 1. 30-78.32) [31,35].
Furthermore, having a family history of PDAC or another hereditary risk may pose a threat. Capurso et al. (2013) [35] compared 390 patients with IPMN with matched controls and found that 5.5% of the patients with Table 1. Characteristics of IPMN based on cellular morphology (data from surgical series) [28][29][30]

Imaging techniques
Currently, cross-sectional imaging plays a main role in lesion detection and differentiation. MRI (combined with MRCP) is the modality of choice, because of its superiority in cyst differentiation and identification of MPD connectivity, mural nodules, and septation.  Fig. 1. a MRCP-diluted pancreatic duct and Santorini with distal a diameter of less than 1 cm. Also, the image of multifocal small sidebranch IPMN. b MRI-ductus pancreaticus which is irregular at the level of the corpus and tail and is slightly dilated. Multiple cysteine deviations starting from the side duct. Largest cystic lesion located in the corpus with a staining solid component. Image matching a mixed-type IPMN with solid component as sign of a possible malignant degeneracy. PA-after pancreatic tail and spleen resection: the ductus pancreaticus and side branches show mixed-type IPMN, both gastric and pancreatobiliary type, with moderate dysplasia; there are extensive regressive changes with mucinous extravasation and fibrosis. No high-grade dysplasia, no malignancy differentiation [42] (Fig. 1). Additionally, the repetitive nature of cyst follow-up mandates a non-invasive modality to eliminate radiation exposure [6•, 10•]. However, for identification of calcifications, tumour staging or surveillance of PDAC recurrence, addition of CT is recommended by some [10•]. Secretin injection during MRCP increases the likelihood of visualising MPD communication, yet only by 5%. More studies are needed to determine whether the addition of secretin outweighs costs and prolongation of scanning time [43].
Endoscopic ultrasound (EUS) is a good alternative for imaging. It is mainly used to assess the presence of worrisome features and should not be performed in case of an established diagnosis or clear indication for surgery. Despite a low accuracy for differentiation between cyst types (61-72%) [44,45], it is highly appropriate for the recognition and delineation of malignant characteristics, especially intracystic structures [46][47][48]. Addition of contrast increases the accuracy of mural nodule detection to 98% [44] (Fig. 2).
An added benefit of EUS is that it allows for cyst fluid collection with fineneedle aspiration (FNA), which is indicated in case of indefinite imaging findings [6•, 7•, 10•]. The AGA recommends EUS-FNA in patients with a cyst diameter ≥ 3 cm, solid component or dilated MPD [8•]. The Fukuoka guideline discourages FNA in case of either high-risk or worrisome features, out of fear for tumour spill [6•]. Cytological cyst fluid analysis has a high specificity (91%), yet low sensitivity (65%) for differentiation between benign and malignant IPMN [45,[49][50][51]. Sensitivity may be increased if the cyst wall and solid components are also sampled [54]. The risk of complications related to cyst EUS-FNA is low (0-2.5%), although higher than for solid lesions. Potential complications are abdominal pain, bacteraemia/infection, haemorrhage and pancreatitis. Prophylactic antibiotics are recommended [51,[55][56][57][58][59] Cyst fluid analysis and biomarkers A broad spectrum of tumour-specific (e.g. mutated KRAS and P53) and tumourassociated (e.g. CA 19-9) markers have the potential to distinguish high-from a b Fig. 2. a and b EUS screenshot captured from D2, the PD is continued from the papilla: focalized dilation over a short trajectory with a diameter of 6 mm, slendering distally with a diameter of 2.7 mm. There is a homogeneous 10-mm cystic lesion not far from the papilla with a connection to de PD. No murine nodule or wall thickening. Conclusion: mixed-type IPMN in pancreatic head and uncinate process low-risk lesions and guide decision-making (Table 2) [10•]. A perfect biomarker should be detectable in an early stage and specific for pancreas neoplasia. Apart from cyst fluid, other potential biomarker sources are serum and pancreatic juice. Glycoproteins are often used as tumour markers. An increased serum level of CA19-9 (9 37 U/ml) is found in 85% of the patients with PDAC and is used to follow the disease course [68]. For IPMN, it is an independent predictor of malignant transformation, with a (pooled) sensitivity  and specificity of 40 and 89%, respectively [69,70]. An increased serum CA19-9 level is a relative indication for surgery and supplementary diagnostics are recommended [10]. Cyst fluid CA19-9 levels have limited clinical value for the identification of advanced neoplastic disease, yet low CA19-9 levels (G 37 U/ml) are suggestive for a non-mucinous origin [51]. Cyst fluid CEA is mainly used for cyst differentiation. A level of G 5 mg/mL is highly specific (95%) for a non-mucinous cyst and a value 9 800 ng/mL for a mucinous cyst (95%) [49]. Little is known about glycoprotein detection in pancreatic juice. Hirono et al. (2012) [58] found a high accuracy (92%) for differentiation between benign and malignant IPMN, based on CEA levels in pancreatic juice (cut-off value 9 30 ng/mL) [58]. Mutated genes are released after cell death and have high potential to serve as biomarkers. Tissue GNAS mutations are associated with IPMN (58-79%; OR 30, 95% CI 7.143-127.622), IPMN-associated adenocarcinoma (36%) and mucinous carcinoma (78%) [71][72][73][74]. In contrast, it is rarely detected in PDAC, PanIn-lesions and MCNs. The prevalence of GNAS mutations differs per morphological subtype: 100% in the intestinal type, 71% in the pancreatobiliary type, 51% in the gastric type and 0% in the oncocytic-type IPMN [75].
KRAS is the driver mutation in most pancreatic PDACs and is also detected in IPMN tissue (50%; OR 7.4, 95% CI 3.9-14.4) [74,76]. However, it is less specific than GNAS, since KRAS is found in 69% of IPMN, 21% of MCN, 90% of PanIn-1 and 90% of PDAC patients [74]. The presence of tissue KRAS and GNAS gene mutations is not related to IPMN location (BD-IPMN vs. MD-IPMN) [74]. In serum, Berger et al. (2016) [77] found that total circulating cell-free DNA levels of 9 0.208 ng/uL distinguish between IPMN and healthy controls with 81% sensitivity and 84% specificity, and between PDAC and healthy controls with 83% sensitivity and 92% specificity. More specifically for GNAS and KRAS, 71% of patients with IPMN harboured cell-free circulating mutated GNAS. Mutated KRAS was not detected in patients with IPMN, although it is present in 42% of patients with PDAC [77]. Adding molecular testing to clinical features and morphology increases sensitivity of IPMN and MCN differentiation to 90 and 94%, respectively. However, more research is needed to distinguish whether the clinical value outweighs the high costs of these sensitive laboratory techniques [59,78]. For pancreatic juice, Suenaga et al. (2018) [60] found GNAS gene mutations in 70% of patients with IPMN. Also, TP53 and SMAD-4 levels were found to be related to dysplasia grade, and able to distinguish IPMN from PDAC with a sensitivity and specificity of 32 and 100%, respectively [60,79]. A VHL gene mutation increases the probability of detecting a serous cyst neoplasia (SCN) [60,79].

Other techniques
Pancreatoscopy uses a thin scope that is introduced in the MPD during ERCP or surgery. It enables intraductal visualisation and image-guided tissue sampling. For differentiation between benign and malignant MD-IPMN, the accuracy is relatively high (88%), yet also are the rates of post-procedural pancreatitis (7%) [80]. During surgery, pancreatoscopy may be combined with intraductal frozen biopsies, to assess the extent of MPD involvement and guide resection [10•, 81].
Needle-based confocal laser endomicroscopy (nCLE) uses a small probe (0.85 mm) that is placed in a pancreatic cyst via a 19-gauge FNA needle and provides a real-time microscopic view (width 320 μm, resolution 3.5 μm). It is able to detect a pancreatic cystic neoplasm with a sensitivity of 59-80% and a specificity of 100%. However, it is currently discouraged by the EU guidelines due to high adverse event rates (7-9%) [10•, [82][83][84][85].  (Table 3). Incentive large-scale prospective registries of individuals undergoing cyst surveillance (e.g. PACYFIC-registry; www.pacyfic.net) are needed to accumulate unbiased data and develop evidence-based guidelines.

Clinical strategy and surveillance
According to all guidelines, the presence of mural nodules or solid components is most predictive for malignant disease. Mural nodules are present in 36-70% of IPMN patients with invasive disease and the size of the mural nodule is correlated with the risk of malignancy [13,20,31,86]. Additionally, a thickened cyst wall is present in~65% of patients with invasive disease (OR 4.80; 95% CI 1.16-14.36) [13,87]. In case of doubt, contrast-harmonic endoscopic ultrasound (CH-EUS) helps to differentiate between mucin and a solid component by the presence of small blood vessels in the latter.
Although cyst size is associated with invasiveness, treatment should not be determined by size alone, since small cysts do not exclude invasiveness and large cysts do not always harbour malignancy [18,19,[88][89][90]. The surveillance intervals in both Fukuoka and ACG guidelines are based on cyst size in the absence of a more practical surrogate [6•, 7•]. The cyst growth appears to be more predictive. A growth of 9 2 mm/year is related to a 45% 5-year risk of developing malignancy versus 1.8% in slowly growing cysts [96][97][98]. Due to a recorded size difference between the different imaging modalities, it is recommended not to alternate modalities between follow-up visits [7•, 10•, 87,94].
The mean MPD diameter is significantly larger in patients with malignant disease. Some guidelines use a 10-mm cut-off value, as absolute indication for surgery [6•, 10•]. This is disputable, since the risk of malignancy is already increased to 59% for patients with a pancreatic duct width between 5 and 9 mm [22]. The AGA and ACG guidelines recommend EUS-FNA in cysts associated with a dilated MPD (ACG cut-off 9 5 mm, AGA non-specified) [  According to the EU, Fukuoka and ACG guidelines, the duration of surveillance should be lifelong. The AGA guideline recommends stopping surveillance in the case of a stable cyst after 5 years. Interestingly, Kwong et al. (2016) [96] found an eightfold higher mortality from non-pancreatic causes than from pancreatic cancer after 5 years of surveillance in low-risk BD-IPMN. On the other hand, multiple studies detected high-risk features in asymptomatic BD-IPMN patients after a follow-up period of more than 5 years [97][98][99]. Additionally, Del Chiaro et al. (2017) [100] found an IPMN-related mortality of 5.8% after 10 years of follow-up in patients without high-risk features at baseline.
After resection of IPMN, lifelong surveillance is recommended, as long as the patient is able and willing to undergo surgery [6•, 7•, 8•, 10•]. He et al. (2013) [101] estimated the chance of developing a new lesion after resection of noninvasive IPMN at 1.6% after 1 year, 14% after 5 years and 18% after 10 years and the chance of invasive pancreatic cancer~0% after 1 year, 7% after 5 years and 38% after 10 years. For invasive IPMN, post-resection surveillance is recommended solely based on symptoms, similar to pancreatic cancer [6•, 10•]. However, one could argue that surveillance should restart (e.g. after~five years) for patients with early-stage invasive IPMN, surveillance should restart after~5 years of survival.
Additionally, data about the incidence of extra-pancreatic neoplasms in patients with IPMN remains controversial, since some retrospective studies show an increased risk in other cancers (e.g. colorectal and gastric cancer) [102][103][104][105]. A large study of 1340 patients by Marchegiani et al. (2015) [36] did not find a higher incidence of extra-pancreatic neoplasms in patients with IPMN. Guidelines do not recommend additional imaging (e.g. CT) for surveillance of extrapancreatic malignancies in patients with IPMN [6•, 7•, 8•, 10•].

Treatment
Guidelines recommend that surgery should be performed by an experienced surgeon in a high-volume centre after consultation and joint decision by a multidisciplinary group with pancreatic expertise. Especially, advanced age and the presence of co-morbidity are related to postoperative mortality of non-pancreatic cause [106][107][108]. On the other hand, early surgery could be considered in younger patients with no comorbidity [9•, 10•]. EU, European; ACG, American College of Gastroenterology; AGA, American Gastroenterological Association; MRI, magnetic resonance imaging; CT, computer tomography; EUS, endoscopic ultrasound; FNA, fine-needle aspiration; CH-EUS, contrast-enhanced harmonic EUS; CA 19-9, cancer antigen 19-9; DM, diabetes mellitus; FU, follow-up; PDAC, pancreatic ductal adenocarcinoma; LGD, low-grade dysplasia; HGD, high-grade dysplasia *To identify calcifications, for tumour staging or for surveillance of recurrence in case of PDAC MD-IPMN and MT-IPMN justify a more aggressive treatment approach than BD-IPMN. In general, surgery should be offered as this is justified by the high prevalence of invasive disease (MD-IPMN 11-81%; MT 20-65%) and the high disease-specific mortality (23 per 1000 patient years; 95% CI 12-52) for untreated MD-IPMN and MT-IPMN [109].
For BD-IPMN, the guidelines are inconsistent and compared in Table 3. The Fukuoka guidelines recommend surgery in the case of ≥ 1 high-risk stigmata or ≥ 1 worrisome features and one of the following: mural nodule ≥ 5 mm, suspicious MPD, suspicious cytology [6•]. The EU guideline is similar, yet in the case of surgical indication, age and the presence of co-morbidity are advised to be taken into account [10•]. ACG stresses the need of decision-making by a multidisciplinary pancreatic group [7•].
In case of suspected malignancy, an oncological resection should be performed. For all IPMNs, intraoperative frozen section examination of the resection margins is recommended. For patients with MD-IPMN or MT-IPMN, intraoperative pancreatoscopy with frozen section of intraductal biopsies can be considered [10•]. Patients with positive margins have a worse survival and extended resection is recommended [15]. Cysts in multifocal IPMNs should be approached autonomously due to their distinct behaviour; the most suspicious lesion(s) should be removed. A total pancreatectomy is only recommended in the case of multiple worrisome features throughout the pancreas or post-surgical recurrence in the remnant pancreas and is performed in 3-37% of the patients. Severe weight loss, diarrhoea (exocrine insufficiency) and/or hypoglycaemic episodes (i.e. brittle diabetes; endocrine insufficiency) are regular consequences of total pancreatectomy [116 117]. However, the majority experiences severe weight loss, diarrhoea (exocrine insufficiency) and/or hypoglycaemic episodes in relation to brittle diabetes (endocrine insufficiency) [110,111]. The survival rates of total pancreatectomy after 1 and 3 years are 80 and 65%, respectively [111].

Prognosis Recurrence after surgery
The overall recurrence rate for IPMN is~11-20% (median 58-73 months), which increases to 65% in the case of malignant IPMN [7•,24,114,115]. For BD-IPMN,~40% is multifocal, which may explain the frequent early recurrence of IPMN in the remnant pancreas (12.5%; mean follow-up 28 months) [116]. Additionally, an increased age, BMI, number of resected lesions as well as an initial location in the pancreatic tail, invasiveness and a family history of PDAC are predictors of recurrence or disease progression [117,118]. The estimated chance to develop a new primary IPMN and related invasive pancreatic cancer after 5 years is 14 and 7%, respectively [101,114,119]. The recurrence rate for MD-IPMN is higher than for BD-IPMN. The dysplasia grade in the resection specimen is the most important predictor of the (severity of) recurrence [24,114,120].

Survival
A large observational study by Marchegiani et al. (2015) [114] found a 5-year survival after resection of 77% for all IPMNs, 69% for MD-IPMN and 82% for BD-IPMN, with a median time to survival of 17, 13 and 24 months, respectively. Vanella et al. (2018) [109] performed a meta-analysis and found a diseasespecific mortality of 23 for all IPMN, 32 for MD-IPMN and 5 for BD-IPMN per 1000 patient years.
In case of invasiveness the overall survival decreases significantly (95% vs. 49%) [114]. Low-grade dysplasia exhibits a similar survival as high-grade dysplasia. In the case of invasive disease, the survival is significantly lower. Of all patients with IPMN-associated adenocarcinoma, 53% has lymph-node metastases, 58% peri-neural and 33% vascular invasion [114,121].

Conflict of Interest
Djuna Cahen is a consultant for Tramedico. Marco Bruno reports grants and personal fees from Boston Scientific, Cook Medical, Pentax, and 3M, outside the submitted work. Iris Levink declares no conflict of interest.

Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by any of the authors.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.