The prevalence of cancer in cases of branch duct intraductal papillary mucinous neoplasm (BD-IPMN) ranges from 6% to 46%.^1–4 The treatment strategy for main duct IPMN is straightforward, and surgical resection is recommended due to the higher malignant potential of this condition.^1,3–9 However, the treatment approach for BD-IPMN is often determined on a case-by-case basis. International consensus guidelines have proposed an algorithm for the surgical treatment of BD-IPMN, which takes into account factors such as cyst size, patient symptoms, and a high-risk appearance (including mural nodules, a thickened, enhancing cyst wall, or positive cytology).^8,9 Since then, several series validating the safety of this approach have been published.^{4,6,7,10–16}

Most published reports concerning the predictive factors of cancer in BD-IPMN have been based on the pathologic results of surgically resected specimens.^{4–7,10–12,14–16} Although surgical pathology is the gold standard of cancer identification, it requires an aggressive surgical approach. A substantial proportion of patients with BD-IPMN can be considered poor surgical candidates due to their advanced age and accompanying comorbidities. While the new consensus guidelines for IPMN⁹ are regarded as an improvement on the 2006 version, questions remain regarding the proper treatment of BD-IPMN and the relative strength of each of these parameters in predicting cancer.

The purpose of this study was to identify clinico-radiologic predictors of disease progression (PD), mainly defined by new-onset symptoms, an interval increase in cyst size, and alterations in morphology, in BD-IPMN patients with relatively long-term follow-up.

This study was conducted in accordance with the principles of the Declaration of Helsinki. Our study protocol obtained approval from the Ethics Committee of Kangbuk Samsung Hospital (KBC11008, approved on February 14, 2011). And for this study, the requirement to obtain informed consent was exempted by the Ethics Committee of Kangbuk Samsung Hospital. All patients (n = 295) with the clinical diagnosis of BD-IPMN treated at our institution from July 2005 to October 2015 were identified from the institutional IPMN database. Of these, 160 patients were excluded due to loss to follow-up (n = 125) or inadequate initial imaging modalities for the diagnosis of BD-IPMN (mostly using only computed tomography [CT] scanning; n = 35). The remaining 135 patients constituted the study cohort and included patients who finally underwent surgical resection with histopathologic confirmation of BD-IPMN (n = 17, 12.6%), and patients who were followed up for presumed BD-IPMN (n = 118, 87.4%) (Fig. 1). Data on patient demographics and clinico-pathologic characteristics were collected by reviewing the medical records.

Figure 1. Flowchart and management of 295 patients with BD-IPMN. BD-IPMN, branch duct intraductal papillary mucinous neoplasm.

Study population

Analyses were done for the 135 patients with BD-IPMN in whom a communication between the cystic lesion and pancreatic duct was confirmed by endoscopic ultrasonography (EUS), magnetic resonance cholangiopancreatography (MRCP), and/or endoscopic retrograde cholangiopancreatography. The BD type was diagnosed in the presence of one or several dilated branches of the pancreatic duct without dilatation of the main pancreatic duct. Unifocal non-resected cystic lesions were required to be ≥ 10 mm, whereas multifocal or pathologically confirmed BD-IPMNs were included independent of size. Unresected cystic lesions were classified as likely BD-IPMN according to the following criteria: endoscopically and/or radiologically proven ductal communication, cyst fluid analysis showing elevated carcinoembryonic antigen (≥ 200 ng/mL), consistent cytology, and/or multifocality. A main pancreatic duct measuring more than 5 mm was regarded as dilated. All enrolled patients were initially unsuitable candidates for surgery due to improper surgical indications (cyst size < 30 mm, no mural nodules, no enhancing wall thickening, or no symptoms) or unsuitable for general anesthesia due to comorbidities, or they refused permission for surgery. We excluded patients with potential causes of chronic pancreatitis, which could be misdiagnosed as IPMN, especially excessive daily alcohol consumption. Patients with a follow-up < 1 year were also excluded. Follow-up imaging, including EUS in combination with MRCP and/or multi-detector row CT was performed once or twice a year. Serial changes in maximum cystic diameter and for high-risk stigmata and worrisome features regarding the risk of malignant transformation according to the revised international consensus guidelines for management of IPMN and mucinous cystic neoplasms of the pancreas,⁹ such as new appearances of mural nodules or enhancing wall thickening, were checked during the study period, which ranged from 12 to 159 months.

Indications for surgery and histological assessment

Surgical resection was recommended according to the Sendai guidelines released in 2006⁸ in the majority of cases. In cases prior to 2006, different surgical indications were applied to some patients. However, no patients underwent surgery in this period. Histological assessments for resected surgical specimens were performed according to the World Health Organization criteria for IPMN¹⁷ by experienced pathologists. The term “malignant transformation” was used only for invasive carcinoma (not for carcinoma in situ), and not for low-, moderate-, and high-grade dysplasia.

Definitions

PD was defined as an interval increase in cyst size > 5 mm in the greatest dimension at any time during follow-up, newly appearing mural nodules or cyst wall thickening, newly appearing focal or diffuse main pancreatic duct dilatation > 5 mm, and newly appearing clinical symptoms and signs. Cancerous changes were defined as identification of intraductal papillary mucinous carcinoma by surgical pathology and/or endoscopic ultrasound guided fine needle aspiration or positive imaging results (interval increase in size and newly appearing metastatic lesions) compatible with malignant disease during clinical follow-up. The follow-up period was defined as the time between the diagnosis of BD-IPMN and one of the following end points: the first imaging showing signs of malignant change, surgical resection of the pancreas, or the last imaging in patients with an uneventful follow-up evaluation.

Statistical analyses

Results are reported as mean ± standard deviation (SD). Continuous variables were compared using the independent-samples t-test for variables showing a Gaussian distribution and the Mann-Whitney U-test for variables not showing a Gaussian distribution. Categorical variables were analyzed using the Fisher exact test or linear-by-linear association test. Cumulative rates for malignant transformation were evaluated using the Kaplan-Meier method with log-rank comparisons. Factors associated with malignant transformation in the univariate analysis were included in a multivariate analysis using Cox forward stepwise linear regression. A P-value < 0.05 was considered statistically significant. Data were collected in Excel 2010 (Microsoft Corp.) and analyzed using SPSS version 13.0 (SPSS Inc.).

Patient characteristics associated with PD

The baseline epidemiological and clinical characteristics are summarized in Table 1. During a mean ± SD follow-up period of 59.4 ± 29.2 months, PD was observed in 28 of 135 (20.7%) BD-IPMN patients. Among these 135 patients, 8 (5.9%) displayed malignant transformation, which was confirmed by surgical pathology in all 8 patients. The remaining 127 patients were categorized into the non-cancerous BD-IPMN group. In this group, 9 (7.1%) patients were confirmed to have non-cancerous BD-IPMN by surgical pathology, while the remaining 118 (92.9%) patients were categorized into this group based on clinical follow-up results (Fig. 2). Patients with PD showed significantly higher proportions with a baseline cyst size > 30 mm, the presence of cyst wall thickening (> 2 mm), and mural nodules than non-PD patients. Pathology for surgically resected specimens also revealed a higher frequency of invasive carcinoma in the PD group than in the non-PD group (Table 2). One case of surgery in the non-PD group involved a patient with cyst wall thickening and mural nodule at baseline who refused surgery initially but consented to surgery during the clinical follow-up.

Table 1 . Baseline Epidemiological and Clinical Characteristics of Enrolled Patients.

Characteristic	n = 135
Age (yr)	62.9 ± 12.8
Sex, male	58 (43.0)
Follow-up period (mo)	59.4 ± 29.2
Baseline diagnostic imaging modalities
MDCT + EUS + MRCP	128 (94.8)
MDCT + EUS	3 (2.2)
MDCT + EUS + ERCP	2 (1.5)
MDCT + EUS + MRCP + ERCP	2 (1.5)
Location
Head/uncinate process	57 (42.2)
Body	45 (33.3)
Tail	11 (8.1)
Multiple	22 (16.3)
Cyst morphology
Unilocular	84 (62.2)
Septate and multilocular	51 (37.8)
Baseline cyst size (mm)	19.1 ± 9.0
Baseline presence of cyst wall thickening (> 2 mm)	8 (5.9)
Baseline presence of a mural nodule	13 (9.6)

Table 2 . Comparisons of the Characteristics between the Patients with PD and Those without PD.

Characteristic	Patients with PD (n = 28)	Patients without PD (n = 107)	P-value
Age (yr)	60.6 ± 13.7	63.4 ± 12.5	0.289
Sex, male	12 (42.9)	46 (43.0)	0.582
Baseline diagnostic imaging modalities			0.801
MDCT + EUS + MRCP	28 (100)	100 (93.5)
MDCT + EUS	0 (0.0)	3 (2.8)
MDCT + EUS + ERCP	0 (0.0)	2 (1.9)
MDCT + EUS + MRCP + ERCP	0 (0.0)	2 (1.9)
Baseline serum CA 19-9 level (U/mL)	13.8 ± 10.8	14.8 ± 9.9	0.676
Baseline serum CEA level (ng/mL)	1.76 ± 0.83	1.72 ± 0.89	0.870
Location			0.935
Head/uncinate process	13 (46.4)	44 (41.1)
Body	7 (25.0)	38 (35.5)
Tail	3 (10.7)	8 (7.5)
Multiple	5 (17.9)	17 (15.9)
Cyst morphology			0.199
Unilocular	15 (53.6)	69 (64.5)
Septate and multilocular	13 (46.4)	38 (35.5)
Baseline cyst size (mm)	22.7 ± 11.6	18.2 ± 8.0	0.060
Baseline cyst size > 30 mm	7 (25.0)	8 (7.5)	0.016
Baseline presence of cyst wall thickening (> 2 mm)	8 (28.6)	0 (0.0)	< 0.01
Baseline presence of mural nodule	12 (42.9)	1 (0.9)	< 0.01
Baseline presence of symptoms	1 (3.6)	0 (0.0)	0.207
Concomitant cancer at another site	6 (21.4)	12 (11.1)	0.208
Cancerous change	8 (28.6)	0 (0.0)	< 0.01

Figure 2. Flowchart of the confirmation methods for 135 patients with BD-IPMN classified by presence or absence of PD. BD-IPMN, branch duct intraductal papillary mucinous neoplasm; PD, disease progression.

Patient characteristics associated with malignant transformation

Univariate analyses revealed that tumor location in the head and uncinate process, septate/multilocular cyst morphology, baseline cyst size ≥ 30 mm, an interval increase in cyst size, baseline cyst wall thickening ≥ 2 mm, baseline presence of mural nodules, and concomitant cancer at another site were significant predictive factors for malignant transformation in BD-IPMN patients (Fig. 3). Using the clinical parameters that were significant in the univariate analysis and adopting age and sex as covariates for multivariate analysis using Cox forward stepwise linear regression, mural nodule (odds ratio [OR] = 58.210; 95% confidence interval [CI], 6.649–509.594; P < 0.01) and concomitant cancer at another site (OR = 8.463; 95% CI, 1.745–41.039; P < 0.01) were significant and independent predictive factors for malignant transformation in BD-IPMN patients (Table 3).

Table 3 . Multivariate Analysis to Identify Independent Contributors to Malignant Transformation in Patients with BD-IPMN.

Variable	Category	Odds ratio	95% confidence interval	P-value
Mural nodules	Baseline presence of a mural nodule (≥ 5 mm)	58.210	6.649–509.594	< 0.01
Concomitant cancer at another site	Presence of concomitant cancer at another site	8.463	1.745–41.039	< 0.01

Figure 3. Kaplan-Meier analyses with log rank comparisons, showing that tumor location in the head and uncinate process, septate/multilocular cyst morphology, baseline cyst size ≥ 30 mm, an interval increase in cyst size, baseline cyst wall thickening ≥ 2 mm, baseline presence of mural nodules, and concomitant cancer of other site were significant predictive factors for malignant transformation in BD-IPMN patients. BD-IPMN, branch duct intraductal papillary mucinous neoplasm.

Concomitant cancers at another site and ductal adenocarcinoma in patients with BD-IPMN

The development of pancreas ductal adenocarcinoma separate from BD-IPMN was evident in 2 of 135 (1.5%) BD-IPMN patients. The predictive clinical factors for malignant transformation in BD-IPMN patients were not associated with the development of pancreas ductal adenocarcinoma. Synchronous or metachronous occurrences of malignant disease in extra-pancreatic organs were noted in 18 (13.3%) patients. The most common malignancy was early gastric cancer (n = 5, 3.7%), followed by thyroid cancer (n = 3, 2.2%), advanced gastric cancer (n = 2, 1.5%), renal cell carcinoma (n = 2, 1.5%), rectal cancer (n = 2, 1.5%), hepatocellular carcinoma (n = 1, 0.7%), breast cancer (n = 1, 0.7%), and hepatocellular carcinoma with concomitant tongue cancer (n = 1, 0.7%) (Table 4).

Table 4 . Concomitant Cancer at Another Site (Non-Pancreatic Malignancies) in the Enrolled Patients with BD-IPMN.

Non-pancreatic malignancies	Previous	Post-diagnosis	Total
EGC	2	3	5
AGC	2	1	3
Breast cancer	1	0	1
Hepatocellular carcinoma	1	0	1
Hepatocellular carcinoma + tongue cancer	0	1	1
Renal cell carcinoma	1	1	2
Rectal cancer	1	1	2
Thyroid cancer	2	1	3
Total patients with malignancy	10 (7.4)	8 (5.9)	18 (13.3)

This study reports on the baseline predictive factors of malignant transformation in patients with BD-IPMN detected by an interval increase in cyst size and alterations in morphology, with relatively long-term follow-up. PD was noted in 28 (20.7%) of 135 BD-IPMN patients. Eight of these 135 (5.9%) patients displayed malignant transformation. A previous report stated that only a minority of patients (20%) required surgery at the time of diagnosis, and of those managed with observation over a median follow-up of 5 years, 21% required surgery.¹⁸ The indications for surgery in that report were the same as in ours (the Sendai 2006 criteria), and the frequency of surgery (around 30%) was similar between our data and that of a previous study. However, in the same study, invasive cancer was present in only 10% of resected cases (of which two-thirds were stage 1a or 1b), and high-grade dysplasia was present in 14% of patients. The reported prevalence of cancer in BD-IPMN varies from 6% to 48%.^1–4 The prior and present differences in the prevalence of cancer in patients with BD-IPMN can be attributed to multiple clinical factors, including differences in region and ethnicity, selection bias via including patients with good compliance, referral center bias, different individual indications for surgical resection in enrolled BD-IPMN patients, and most importantly, differences in baseline and follow-up imaging modalities. EUS and/or MRCP were universally adopted in the current study to confirm the communication between the cystic lesion and pancreatic duct at baseline and new appearance in the follow-up for worrisome features and stigmata of malignancy, whereas other studies^1–4 adopted various imaging modalities, which could not always confirm communications between the cyst and pancreatic duct.

The 2012 Sendai guidelines were formulated based on clinical studies; hence, the phrase “evidence-based consensus” was added. However, there are still areas of uncertainty, most salient in the criteria for resection of BD-IPMN. Most criteria for surgery in BD-IPMN are based on morphological features deemed to be associated with a risk of cancer identified by various imaging modalities. Recent advances in imaging modalities have led to the incidental identification of many asymptomatic cystic lesions in the pancreas, and preoperative cytopathological diagnosis are not always possible or sufficient to determine surgical indications for these patients. EUS is more sensitive in discriminating benign and malignant IPMN than other imaging modalities, such as CT or magnetic resonance imaging (MRI). EUS is also very useful for identifying small tumor lesions in BD-IPMN, such as mural nodules.^19–22 In the current study, baseline clinical parameters as a predictor of malignant transformation in patients with BD-IPMN included tumor location in the head and uncinate process, septate/multilocular cyst morphology, baseline cyst size ≥ 30 mm, an interval increase in cyst size, baseline cyst wall thickening ≥ 2 mm, baseline presence of mural nodules, and concomitant cancer at another site in the univariate analysis (Fig. 3). However, the independent predictive factors of malignant transformation in patients with BD-IPMN by Cox forward stepwise linear regression analysis were only a mural nodule and concomitant cancer at another site (Table 3). The diagnostic discrimination values of EUS for detecting the predictive factors of malignant transformation were reported to be higher than those of other diagnostic modalities such as CT and MRI.^19–22 Hence, the adoption of EUS for diagnostic imaging in the routine follow-up of patients with BD-IPMN could be considered to be reasonable. However, the invasive nature and significant intra- and inter-observer variability could be obstacles to the routine application of EUS in follow-up.

Among the radiologic predictors of malignancy included in the 2012 consensus guidelines, a thickened cyst wall was identified as a significant clinical parameter predicting malignant transformation in patients with BD-IPMN in the current study. Although few reports have evaluated a thickened cyst wall as a predictive factor for malignancy in patients with BD-IPMN, it has proven consistently valuable.²³ Nonetheless, this radiologic parameter had a sensitivity of only 3%–7%, despite its high positive predictive value, indicating that this parameter may be useful in the diagnosis of malignant transformation in BD-IPMN patients.

Mural nodules are the most suggestive imaging finding for malignant BD-IPMN. Many previous reports and a recently published meta-analysis have underscored the importance of mural nodules in predicting malignancy in patients with BD-IPMN.^{13–15,18–23} The presence of a mural nodule was also a significant and independent predictor of malignant transformation in patients with BD-IPMN in the current study. However, the enhancement pattern of mural nodules using intravenous contrast administration was not checked during EUS examinations in this study, which made it impossible to confirm whether each mural nodule was a true neoplastic growth. Given the possibility of some non-neoplastic intra-cystic debris or mucin drops presenting as a mural nodule in the collected data, the diagnostic value of mural nodules as an indicator of malignant transformation might be even greater if true enhancing mural nodules are considered separately. Nevertheless, in the present study, the presence of a mural nodule, regardless of its enhancement, increased the possibility of malignancy by about 58-fold. The present results support the idea that it may be acceptable to consider the presence of a thickened cyst wall and/or a mural nodule as a highly suspicious finding for malignancy that would warrant recommending surgical resection.

Ductal adenocarcinoma of the pancreas separate from BD-IPMN developed in 2 of 135 (1.3%) BD-IPMN patients during the follow-up period. This incidence is lower than the previously reported 6.9% 5-year rate of development of ductal carcinoma in BD-IPMN patients.²⁴ Interestingly, the endoscopic and/or radiologic parameters significantly associated with malignant transformation in BD-IPMN patients did not have any significant associations with the development of ductal carcinoma in the same patients. Different biological backgrounds between the malignant transformation of BD-IPMN and ductal carcinoma not associated with BD-IPMN might be an explanation.

Synchronous or metachronous occurrence of malignant diseases in extra-pancreatic organs has been reported to occur in 20%–30% of patients with IPMN.²⁵ The frequency of all extra-pancreatic malignancies in the present study was 13.3% (18/135). In addition, the metachronous malignancy rate was 5.9% (8/135). Therefore, attention should be paid to this phenomenon, and careful surveillance should be conducted even after the resection of IPMN lesions. Screening for colorectal polyps and cancer is recommended in the United States.^26,27 However, the present results indicate that surveillance for gastric cancer (early and advanced) should be emphasized in East Asia, a region where gastric cancer is highest. Concomitant cancer at another site was a significant and independent predictor of malignant transformation in patients with BD-IPMN in the current study. A recent study²⁸ also reported that six types of extra-pancreatic malignancies occurred in 15 out of 51 histologically confirmed IPMN patients (29.4%). The standardized incidence ratio (SIR) for extra-pancreatic malignancies was significantly higher in IPMN patients than in healthy controls (SIR = 2.18; 95% CI, 1.31–3.42; P = 0.004). Notably, benign IPMNs showed no association with extra-pancreatic malignancies (SIR = 0.92; 95% CI, 0.43–1,76; P = 0.87). In contrast, malignant IPMNs showed a higher association (SIR = 3.83; 95% CI, 1.87–7.03; P = 0.0009). Thus, only malignant IPMNs drive the significant association with prior extra-pancreatic malignancies, showing a nearly four-fold increased incidence compared to the general population. We hypothesize that these observations presented in the current study and a recently reported study²⁸ may be explained if some patients with extra-pancreatic malignancies presented a higher risk of developing IPMNs (and vice versa), possibly resulting from an uncharacterized condition that predisposes them to multiple cancers.

In conclusion, the current study demonstrated that a substantial proportion of patients with BD-IPMN showed PD and malignant transformation during long-term follow-up. Mural nodules and concomitant cancer at another site were significant and independent predictive factors of malignant transformation in patients with BD-IPMN. It can be feasibly recommended to conduct careful endoscopic and radiologic examinations, especially using EUS to detect newly developed mural nodules in high-risk patients who have a previous history or newly diagnosed concomitant cancer at another site during the follow-up monitoring of BD-IPMN.

None.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Hong Joo Kim has been an editor (Deputy Editor) of the International Journal of Gastrointestinal Intervention (IJGII) since 2017; however, Hong Joo Kim has not been involved in the peer reviewer selection, evaluation, or decision process of this article. No other potential conflicts of interest relevant to this article were reported.