Intraductal papillary mucinous neoplasms (IPMNs): is it time to (sometimes) spare the knife?
- Professor Claudio Bassi, Department of Surgery, University of Verona, Piazzale LA Scuro, 10, 37134 Verona, Italy;
A study by Tanno et al1 published in this issue of Gut (see page 10.1136/gut.2007.129684) shows that a branch-duct intraductal papillary mucinous neoplasms of the pancreas (BD-IPMNs) without mural nodules can be safely followed with non-operative management, confirming the results of a prospective study carried out at our institution and recently published in this journal.2
IPMNs were first described 25 years ago by Ohashi and colleagues with the name “mucin secreting cancer of the pancreas”.3 During the 1980s and early 1990s many case reports and small series of IPMNs were published but at that time IPMNs were frequently confused with other pancreatic tumours, in particular with mucinous cystic neoplasms (MCNs) of the pancreas.4 5 It was only in 1996 that the World Health Organization made a clear distinction between IPMNs and MCNs, the latter being characterised by the presence of the unique ovarian stroma with no involvement of the ductal system of the pancreatic gland.6 Mucin-producing neoplasms of the pancreas, including both IPMNs and MCNs, were considered as lesions with potential or actual malignancy, and therefore surgery was recommended for all mucinous tumours.4 5
IPMNs can affect the ductal system of the pancreas with two distinct forms: neoplasms that arise from the main duct with or without a synchronous involvement of the branch ducts (MD-IPMNs) or IPMNs involving only the secondary branches (BD-IPMNs).
The incidence of IPMNs has increased dramatically over the past 10 years and, nowadays, they represent the second indication for pancreatic resection at our institution, after ductal adenocarcinoma. This is basically related to the widespread use of cross-sectional imaging techniques such as computed tomography (CT) and magnetic resonance cholangiopancreatography (MRCP) that led to “incidental” diagnosis of cystic neoplasms of the pancreas (including IPMNs) in a large number of asymptomatic patients.7–9 In particular, the number of radiological diagnoses of small (<2 cm) BD-IPMNs without solid elements has significantly increased and it has been termed a “disease of technology”.10
In the late 1990s some authors showed that BD-IPMNs were not only topographically different but had less-aggressive biological behaviour with a lower rate of malignancy.11 12 Moreover Irie et al and Fukukura et al reported a series of radiological parameters significantly related to malignancy, such as a diameter >3 cm, the presence of nodules, and the thickness of the cyst wall.5 13 14 This, in turn, raised the question of whether some BD-IPMNs could undergo non-operative management instead of surgical resection.15 The importance of this new scenario was evident. In fact, BD-IPMNs are frequently discovered in elderly patients with potential comorbidities that can affect the outcome after pancreatic resections.2 5 7 Most lesions occur in the head of the gland, requiring a pancreaticoduodenectomy as treatment of choice, while those located in the body tail will entail a left pancreatectomy, with or without spleen preservation. It is true that, nowadays, pancreatic resections are safe and can be accomplished with a low mortality (<1–2%), at least in referral centres, but morbidity still remains high with a significant risk of exocrine and endocrine insufficiency.16
Can we spare the knife in selected patients with BD-IPMNs? Is this approach safe and feasible? In this issue of Gut, Tanno et al1 tried to answer this important question. This paper reports on the follow-up of 82 patients with BD-IPMNs without mural nodules, who have been diagnosed and followed with endoscopic ultrasound, CT scan or MRCP performed every 6 or 12 months for a median of 61 months (range 14–148). 98.8% of the patients were asymptomatic and the median size of the cyst at diagnosis was only 20 mm and only 10 patients (12%) had a size greater than 30 mm. Nine patients (11%) showed an increase of the size of the cyst after a median of 59 months; of these only one had symptoms and finally three patients underwent surgical resection with a histopathological diagnosis of adenoma in two cases and of borderline neoplasms in one. Another four patients (5%) developed mural nodules after a median follow-up of 105 months. They were asymptomatic and underwent surgery that revealed the presence of BD-IPMNs with adenoma in three cases and with carcinoma in situ in the remaining one. In conclusion, this study shows that most BD-IPMNs without mural nodules remain unchanged during follow-up, and can be followed with a non-operative management. In those patients who showed cyst enlargement or appearance of mural nodules no invasive cancer was found and this underlines the safety of this approach.
At the moment there are few prospective studies focusing on the conservative management of patients with BD-IPMNs that have a significant number of patients and good follow-up data.17 18 In 2000 we started a prospective study at our institution to follow up asymptomatic patients with a clinico-radiological diagnosis of BD-IPMNs.2 Patients who presented symptoms and/or malignancy-related parameters (size >3.5 cm, nodules, thick walls, CA 19.9 >25 U/l, recent onset or worsening diabetes) underwent surgery while the remaining patients were followed according to a clinico-radiological protocol, based on MRCP with secretin and transabdominal ultrasound with contrast medium. Eighty-nine patients were followed for a median of 32 months; only five of them (5.6%) showed an increase in lesion size and underwent surgery. At pathology three BD-IPMNs with adenoma and two with borderline neoplasms were found. Despite a relatively short follow-up, our data suggested that non-operative follow-up seems to be effective in asymptomatic patients in the absence of malignancy-related parameters.
The protocol adopted at our institution is very similar to the guidelines for the management of IPMNs put forward by the International Association of Pancreatology (IAP), which suggest that BD-IPMNs that cause no symptoms, measure less than 3 cm and have no nodules can be observed with periodic imaging.5 Analysing the combined experience of the University of Verona and the Massachusetts General Hospital with 145 resected, pathologically confirmed BD-IPMNs, there was no case of carcinoma (in situ or invasive) in absence of symptoms unless the lesion was greater than 3 cm in size or had nodules/thick walls,7 confirming the validity of IAP guidelines.
A very important difference must be noted between BD-IPMNs and MD-IPMNs. In fact BD-IPMNs exhibit lower rates of malignancy and invasive carcinoma than MD-IPMNs: in our experience 22% of BD-IPMNs are malignant with a rate of invasive cancer of 11%, while in MD-IPMNs these rates rise to 60% and 42%, respectively.4 7 Similar data are reported by other authors.5 19 Moreover, in MD-IPMNs the age difference between non-invasive and invasive tumours suggests tumour progression.4 19 For these reasons MD-IPMNs must be resected in all surgically fit patients; surgical resection should be routinely associated with the frozen section examination of the main duct resection margin; the presence of any “positive” resection margin should push the surgeon to perform further resections up to total pancreatectomy.5 20 Certainly the decision of whether or not to perform a total pancreatectomy should be balanced with a patient’s age and comorbidities as well as the degree of dysplasia at the resection margin (i.e. adenoma versus carcinoma).21 Thus, preoperatively, a differential diagnosis between MD-IPMNs and BD-IPMNs is of paramount importance: in the presence of a dilated main pancreatic duct the suspect of MD-IPMNs must be carefully considered.
The strengths of the paper by Tanno et al1 are that this is a prospective study with a large number of patients (even if collected over a 16-year period) and with a long-term follow-up. On the other hand this study shows some drawbacks. The presence of nodules is considered the most important parameter influencing the management of patients with BD-IPMNs, while the reader should remember that the combination of several parameters (presence of symptoms including diabetes, size, presence of nodules and thick walls) should be evaluated to establish whether non-operative management can be considered.2 The presence of mural nodules is always a clear indication for resection. In our experience with 145 resected patients with BD-IPMNs the likelihood of malignancy increased significantly with the presence of nodules, which were absent in 66 patients with adenoma and were present in four of 47 borderline neoplasms (8.5%), seven of 16 in situ carcinomas (43.8%) and 12 of 16 invasive carcinomas (75%).7 Symptoms are another important indication for surgery. Patients with BD-IPMNs who have symptoms should undergo surgery not only to alleviate symptoms, but also because they have a higher incidence of malignant tumours.5
IAP guidelines suggest that BD-IPMNs greater than 3 cm in size should undergo surgical resection.5 In the protocol adopted at our institution we chose a cut-off of 3.5 cm, and in our experience an increase in tumour size does not necessarily indicate malignancy or tumour progression.2 A similar observation was made by Tanno et al.1 Maybe the cut-off of 3 cm proposed by the IAP will be increased in the future. However, other prospective studies should confirm these findings and at the moment neoplastic growth must remain an indication for surgery in order to avoid the danger of under-treatment.
Which kind of follow-up should we perform in asymptomatic patients with BD-IPMNs without malignancy-related parameters? Every 6–12 months these patients should undergo a careful clinical examination with CA 19.9 and blood sugar estimation as well as imaging techniques such as CT, MRCP or endo-ulbra sonography (EUS).1 2 5 8 17 22–25 . At initial diagnosis EUS could be followed by a fine-needle aspiration (FNA) of the lesion to look for the presence of atypical cells, mucous and carcinoembryonic antigen (CEA) within the cyst fluid.24 25 In the future it might be possible that the expression of more specific molecular markers could be useful in the differential diagnosis between BD-IPMNs and other cystic neoplasms of the pancreas and, more important, in defining their biological behaviour. Fukushima and co-workers26 found methylated ppENK in the pancreatic juice of patients with IPMN with carcinoma in situ but in none of those with low-grade IPMNs. Fascin over-expression has been found in IPMNs, and a correlation with increased histological grade has been demonstrated.27 The expression of mucins, particularly MUC4 and MUC5AC might be useful in differentiating low-grade and high-grade IPNs as well.28
MRCP with secretin has a high sensitivity and specificity in characterising IPMNs preoperatively, and in our experience it is the best method to characterise IPMNs and outline their gross appearance.2 8 23 MRCP can indicate the site and the extension of main pancreatic duct dilatation, the presence of a communication between the duct and the cyst lesion, the presence and size of mural nodules and thick walls, and of multifocal branch-duct IPMNs.5 8 29
We usually use MRCP with secretin stimulation at initial diagnosis. In fact, this is a dynamic examination and, in our experience, the use of secretin proved to be particularly useful to evaluate the characteristics of the main pancreatic duct and its branch ducts as well as in demonstrating the presence of the communication in BD-IPMNs.2
High-resolution multislice helical CT with dedicated pancreas protocol can provide good details of the morphological characteristics of IPMNs, even if MRCP seems to be more sensitive in the diagnosis of small branch-duct IPMNs, especially in the setting of a multifocal disease.14 23 29 If MRCP does not provide sure diagnosis, more information can be obtained from EUS. Fine-needle aspiration cytology and cyst-fluid analysis for amylase and different tumour markers may provide further differential diagnostic information.17 24 25 The reader should keep in mind, however, that EUS is an operator-dependent and invasive technique and we think that, in the case of non-operative management of IPMNs, it is easier, over time, to compare the different lesions (especially in the case of multifocal BD-IPMNs) with MRCP rather than with EUS.
For years, ERCP has been the procedure of choice for the diagnosis of IPMNs, allowing brushings or aspiration of the main pancreatic duct for cytology.5 Nowadays, the diagnostic value of ERCP is less, due to the significant improvements in radiological imaging and EUS.29 In addition, ERCP carries the risk of inducing acute pancreatitis and in BD-IPMNs it may fail to fill the cystic side branch due to mucus plugging the route.
Finally, an important issue related to BD-IPMNs is represented by multifocality. Unfortunately, in the paper by Tanno et al there is no mention of this. Rodriguez et al demonstrated the presence of multifocal lesions in 37 out of 145 patients (25.5%) who underwent resection for BD-IPMNs;7 multifocality was not significantly different between benign and malignant tumours (24.8% v 28.1%). 64% of 89 patients who underwent a non-operative management at our institution had multifocal lesions as well, and in most of them the entire pancreas was involved.2 The presence of multifocal BD-IPMNs is a real challenge for the clinician; in fact, in cases of surgical approach, extensive pancreatic resections up to total or subtotal pancreatectomy should be considered, while for those patients in whom a non-operative management is feasible, the presence of multiple BD-IPMNs along the gland can be particularly difficult to evaluate over time.
The prognosis of IPMNs even when malignant and invasive can be quite good.2 4 5 19 In a few years the outlook for these neoplasms has significantly changed as a result of increased knowledge in this setting. Nowadays, surgeons can sometimes spare their knife in front of patients with BD-IPMNs. The careful selection of these patients is of paramount importance in order to avoid under-treatment. For this reason we believe that at the moment this non-operative approach should be limited to referral centres with expertise in the field. In the future, molecular biology and genetics will probably contribute to improvements in this “selection process” with advantages for the clinical management of patients with BD-IPMNs.