Incidental cystic lesions of the pancreas (CLPs) are increasingly detected with the improvements in high-resolution abdominal imaging and increasing frequency of the use of this tool.^1,2 Many of these CLPs are proved to be intraductal papillary mucinous neoplasm (IPMN)³ and these entities of CLPs are well recognized to have malignant potential, which often makes clinicians to consider surgical resection. However, given that the potential morbidity and mortality of major pancreatic surgery is high, strategies of surgical resection to all of incidental CLPs do not appear desirable.

Many studies have investigated the natural history of CLPs with knowledge accumulated.^4-14 It is well recognized that some CLPs have a low risk of malignant transformation with a favorable prognosis without surgery. Based on this knowledge, current consensus guidelines recommend surveillance for such CLPs.^15-18

However, treatment and surveillance strategies can vary. They are inconsistent among guidelines. Few studies have addressed a practical method of surveillance to date. An optimal method of surveillance for CLPs such as duration and interval of follow-up has not well established in clinical practice. Our institutions have collected a cohort of CLPs. Patients in this cohort have received follow-up with defined protocol prospectively. The aim of the current study was to investigate long-term clinical outcomes of the cohort and clinical validity of the current consensus guidelines to provide data for determining the optimal duration and interval of surveillance for CLPs in clinical practice.

1. Patients

Patients older than 20 years of age who were diagnosed as having CLPs by abdominal ultrasonography or computed tomography (CT) were enrolled in the cohort in Seoul National University Hospital. Patients with evidence of pancreatitis, a history of von Hippel-Lindau disease, a polycystic disease of the kidney of liver, or cystic fibrosis were excluded from enrollment in this cohort.

Among patients in this cohort, those who were diagnosed as having CLPs from January 2000 to December 2008 were included in the current study. Exclusion criteria were: (1) patients with follow-up period of less than 6 months, and (2) patients having CLPs with worrisome features (WFs) or high-risk stigmata (HRS) according to the revised 2017 International Association of Pancreatology (IAP) Guidelines.¹⁵ The definition of HRS and WFs are provided in Supplementary Materials.

The study protocol was reviewed and approved by the Institutional Review Board of Seoul National University Hospital (IRB number: H-1406-002-582) and the informed consent was waived.

2. Methods of surveillance

All enrolled patients in our cohort received follow-up with CT prospectively according to a defined protocol. Briefly, patients were followed up every 6 months within the first 2 years. They were then followed up every 12 months in case of no change in cystic size or feature during the first 2 years. Measurement of the size of CLP is provided in Supplementary Materials. Cyst growth rate was calculated as difference between size at diagnosis and size at last follow-up divided by the number of years of follow-up.

For each follow-up time, we investigated and documented whether WF, HRS, and surgical indication were developed, whether patient underwent surgery, and whether patients were diagnosed with malignancy after surgery. Surgical indication was defined as the presence of symptoms or complications related to CLPs, the development of WF or HRS, or the development of concomitant pancreatic solid mass presumed malignancy.¹⁵ In patients with development of surgical indication during follow-up, enhanced magnetic resonance imaging with magnetic resonance cholangiopancreatography or endoscopic ultrasonography was performed. Decision to perform surgical resection or further follow-up was made according to clinical judgement in each case by both gastroenterologists and surgeons. The duration of follow-up was defined as a period from the initial detection of CLP to the last imaging follow-up of CLP (in case of surgery, the last imaging follow-up before surgery).

3. Identification of malignancy progression

For cases of surgery, the diagnosis of malignancy was made based on pathological examinations of surgical specimens. For cases without surgery, the diagnosis of malignancy was made based on typical imaging findings with compatible clinical course, in which case, surgery or endoscopic ultrasonography-guided fine needle aspiration was performed according to discretion after multidisciplinary approach. Any pathology with invasive carcinoma and IPMN with high-grade dysplasia were considered as malignancy. Pancreatic carcinoma was considered to be derived from IPMN when it originated within the area with known CLP and when invasive solid mass extended continuously to the CLP. It was considered a concomitant or distinct pancreatic carcinoma when the invasive component was located separately from the CLP.

4. Outcomes

The primary outcome in the current study was defined as development of surgical indication during follow-up. The secondary outcome was defined as surgery and malignant progression during follow-up. We investigated these outcomes using a retrospective analysis with prospectively collected data.

5. Validation of consensus guidelines

In the current study, clinical validation of the 2017 IAP guideline and the 2018 European guideline was performed by investigating diagnostic performances of both guidelines predicting malignant CLPs. The investigation was performed with only patients who underwent surgery. Patients with diagnosis of concomitant carcinoma were not included in this analysis. We evaluated the importance of each risk factor in the category of HRS and WF of the 2017 IAP guideline and in the category of absolute and relative indication of the 2018 European guideline. We then investigated the diagnostic value of risk score defined as the number of positive risk factors in predicting malignant CLPs.

6. Statistical analysis

Differences in categorical variables were analyzed using the chi-square test. Continuous variables expressed as means±standard deviations were compared using the Student t-test. Univariate survival analysis was performed using the Kaplan-Meier method. The log-rank test was used to compare differences. A Cox proportional hazard regression model was used for multivariable analysis. Time from first observation to development of surgical indication was used to assess the risk of developing surgical indication during surveillance. To assess the risk of developing malignancy, time from first observation to development of malignancy was calculated. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated for each risk score. Receiver operating characteristic curve analysis was generated and areas under the curves (AUCs) were statistically compared to evaluate diagnostic values of different risk scoring systems. All statistical analyses were performed with SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Statistical significance was defined at p-value <0.05.

1. Process of initial identification and enrollment of patients

From January 2000 to December 2008, a total of 628 patients were diagnosed with CLPs. Of these patients, 121 (19.3%) patients with prior clinical and laboratory evidence of pancreatitis and 23 (3.7%) patients with polycystic kidney disease or von Hippel-Lindau disease were excluded. Of the remaining 484 patients who were enrolled in our cohort, 72 patients (14.9%) underwent surgical resection within 6 months after initial diagnosis for high suspicion malignancy or presence of symptoms and 163 patients (33.7%) refused to receive follow-up. As a result, 249 patients underwent surveillance. Of them, 22 patients had CLPs with WF, HRS, or typical imaging findings of serous cystic neoplasm. Finally, 227 patients with CLPs without any WF or HRS were included in the current study. The process of initial identification and enrollment of patients for the current study is depicted in Fig. 1.

Figure 1. Algorithm for initial identification and enrollment of patients. CLPs, cystic lesions of the pancreas; PKD, polycystic kidney disease; VHL, von Hippel-Lindau disease; WF, worrisome feature; HRS, high-risk stigmata.

2. Patient demographics and cyst growth

The baseline characteristics of enrolled 227 patients are summarized in Supplementary Table 1. The median age of the enrolled patients was 62 years (range, 34 to 81 years). The median cyst size was 13 mm (range, 2 to 28 mm). The median follow-up period for all enrolled patients was 120 months (range, 6 to 314 months). Two hundred and ten patients (92.5%) underwent surveillance for more than 5 years (median 125 months; range, 60 to 264 months) and 114 (50.2%) were followed up for more than 10 years (median 144 months; range, 120 to 264 months). The pattern of change in the size of CLPs during follow-up for individual cases with natural histories and clinical outcomes of CLPs is shown in Fig. 2. The median annual rate of cyst growth was 0.3 mm (range, –2.0 to 6.3 mm). In patients who underwent surgery, the median growth rate was 1.7 mm (range, –0.6 to 6.3 mm).

Figure 2. Patterns of the change in the size of cystic lesion of the pancreas during follow-up.

3. Surgery and malignancy

Twenty-two patients (9.7%) underwent surgery during the follow-up period. Causes of surgery, type of surgery, and final pathologic results are provided in Table 1. The cause of surgery was the presence of WF or HRS in 15 patients. However, in two patients, a solid mass located separately from CLP developed during follow-up and surgery was performed for a presumptive diagnosis of pancreatic cancer. The median duration of follow-up before surgery was 47.5 months (range, 7 to 134 months). Final pathologic results revealed malignancy in four patients.

Table 1

Surgery and Pathologic Results.

Variable	Data (n=22)
Follow-up till surgery, median (range), mo	48 (7–134)
Indication for surgery, No. (%)
Cyst growth	10 (45.4)
Solid component	8 (36.4)
Main pancreatic duct dilatation	2 (9.1)
Symptoms or complications	1 (4.5)
Patient’s anxiety	2 (9.1)
Concomitant solid mass	2 (9.1)
Simultaneous surgery on another tumor	2 (9.1)
Type of surgery, No. (%)
Pancreaticoduodenectomy	5 (22.7)
Distal pancreatectomy	12 (54.5)
Middle pancreatectomy	3 (13.6)
Subtotal pancreatectomy	2 (9.1)
Pathology, No. (%)
IPMN with low grade dysplasia	14 (63.6)
MCN	1 (4.5)
SCN	3 (13.6)
Carcinoma derived from IPMN	2 (9.1)
Concomitant carcinoma (PDAC, NEC)	2 (9.1)

IPMN, intraductal papillary mucinous neoplasm; MCN, mucinous cystic neoplasm; SCN, serous cystic neoplasm; PDAC, pancreatic ductal adenocarcinoma; NEC, neuroendocrine carcinoma..

Data of four patients with malignancy is summarized in Supplementary Table 2. In two patients, surgery was performed for the development of HRS (solid component) occurred at 62 and 68 months after the initial diagnosis. The final pathology revealed carcinoma derived from IPMN. However, in the other two patients, surgery was performed for the development of solid mass located separately from CLP that occurred at 50 and 109 months of follow-up. The final pathology revealed concomitant pancreatic carcinoma (ductal adenocarcinoma and neuroendocrine carcinoma, respectively). The incidence of malignancy during surveillance was 8.3% (3/36) in patients with CLPs developing WF or HRS and 0.5% (1/191) in patients without WF or HRS.

4. Clinical episodes in each follow-up time

During surveillance, the development of HRS occurred in two patients (0.9%) (solid component at 62 and 68 months of follow-up) and 39 patients (17.2%) experienced the development of WF after a median time of 78 months (range, 15 to 229 months).

To determine the optimal duration and interval of surveillance for CLPs, we investigated major clinical episodes at each follow-up time. The cumulative incidence of major clinical episodes at each time point are summarized in Table 2. The major clinical episodes within 10 years are provided in Supplementary Materials.

Table 2

Cumulative Incidence of Clinical Episodes in Each Follow-up Time.

Follow-up duration	Malignancy	Surgery	Presence of pancreatitis	Presence of WF	Presence of HRS
Up to 6 mo	0	0	0	0	0
Up to 12 mo	0	3 (1.3)	0	0	0
Up to 18 mo	0	4 (1.8)	0	2 (0.9)	0
Up to 24 mo	0	7 (3.1)	0	6 (2.6)	0
Up to 36 mo	0	7 (3.1)	0	8 (3.5)	0
Up to 48 mo	0	11 (4.8)	0	12 (5.3)	0
Up to 60 mo	1 (0.4)	14 (6.2)	1 (0.4)	14 (6.2)	0
Up to 72 mo	3 (1.3)	17 (7.5)	1 (0.4)	18 (7.9)	2 (0.9)
Up to 84 mo	3 (1.3)	19 (8.4)	1 (0.4)	22 (9.7)	2 (0.9)
Up to 96 mo	3 (1.3)	19 (8.4)	1 (0.4)	26 (11.5)	2 (0.9)
Up to 120 mo	4 (1.8)	21 (9.3)	1 (0.4)	28 (12.3)	2 (0.9)
Up to 144 mo	4 (1.8)	22 (9.7)	1 (0.4)	31 (13.7)	2 (0.9)
Up to 180 mo	4 (1.8)	22 (9.7)	1 (0.4)	32 (14.1)	2 (0.9)
Up to 270 mo	4 (1.8)	22 (9.7)	1 (0.4)	39 (17.2)	2 (0.9)

Data are presented as number (%)..

WF, worrisome feature; HRS, high-risk stigmata..

One hundred and fourteen patients (50.2%) were followed up for more than 10 years (median 144 months; range, 120 to 264 months). Of them, 100 patients did not experience the development of WF or HRS within 10 years. While WF was developed after 10 years of follow-up in nine (8.3%) of these 109 patients, no malignancy was developed after 10 years of follow-up.

5. Development of surgical indication during follow-up

We investigated the cumulative rate of progression to surgical indication during follow-up. Overall, 37 (16.3%) patients experienced the progression to surgical indication. The rate of progression to surgical indication was 34.9% (15/43) in patients with initial cyst size more than 2 cm and 12.0% (22/184) in patients with initial cyst size less than 2 cm.

Fig. 3 shows cumulative incidence of progression to surgical indication at each follow-up time. The number in brackets indicates the cumulative number of patients who received actual surgery in each follow-up time. No patients with initial CLP size less than 2 cm experienced the progression of surgical indication within 24 months of follow-up. In patients with initial CLP size more than 2 cm, development of surgical indications did not occur within 12 months. Larger initial cystic size and faster growth rate were found to be independent predictors of progression to surgical indication according to multivariable analysis (Table 3).

Figure 3. Cumulative incidence of the progression to surgical indication. The number in bracket means the cumulative number of patients who received actual surgery in each follow-up time. (A) Overall patients. (B) Stratification with initial cyst size.

Table 3

Univariate and Multivariate Analyses of Predictors for Progression to Surgical Indication.

	Univariate analysis			Multivariate analysis
	HR (95% CI)	p-value		HR (95% CI)	p-value
Sex		0.819
Male	1
Female	0.925 (0.477-1.796)
Age		0.860
<65 yr	1
≥65 yr	0.938 (0.458-1.922)
Initial cystic size		<0.001			<0.001
>2 cm	1			1
≥2 cm	4.166 (2.128-8.156)			6.173 (3.111-12.248)
Growth rate		<0.001			<0.001
<0.5 mm/yr	1			1
≥0.5 mm/yr	10.671 (4.665-24.410)			13.487 (5.838-31.159)

HR, hazard ratio; CI, confidence interval..

6. Validation of the current consensus guidelines

To validate the 2017 IAP guideline and the 2018 European guideline, we investigated diagnostic performances of both guidelines predicting malignant CLPs. Table 4 shows diagnostic values of different risk scores for predicting malignant CLPs. The highest accuracy was observed at WF score of 3 or greater, WF plus HRS score of 3 or greater, and WF plus HRS score of 4 or greater, in which the accuracy was 94.4%. Receiver operating characteristic curve analysis showed that the AUC of WF plus HRS scores was higher than that of WF scores and that the AUC of relative indication plus absolute indication scores was higher than that of relative indication scores (Fig. 4). Among different risk scoring systems, the AUC of WF plus HRS scores was the highest at 0.986 (95% confidence interval, 0.936 to 1.000). HRS scores and absolute indication scores were not used in this analysis because of a small number of theses scores (only 0 or 1).

Figure 4. Receiver operating characteristic curves of different risk scores for predicting malignant cystic lesion of the pancreas.
WF, worrisome feature; HRS, high-risk stigmata, AUC, areas under the curve; CI, confidence interval.

Table 4

Diagnostic Value of Different Risk Scores in Predicting Malignant Cystic Lesions of the Pancreas.

Score	Benign (n=18)	Malignancy (n=2)	p-value	Sensitivity, %	Specificity, %	PPV, %	NPV, %	Accuracy, %
WF score
≥1	13	2	>0.999	100	27.8	13.3	100	27.8
≥2	4	2	0.079	100	77.8	33.3	100	77.8
≥3	1	2	0.016	100	94.4	66.7	100	94.4
≥4	1	1	0.195	50	94.4	50	94.4	47.2
WF+HRS score
≥1	13	2	>0.999	100	27.8	13.3	100	27.8
≥2	4	2	0.079	100	77.8	33.3	100	77.8
≥3	1	2	0.016	100	94.4	66.7	100	94.4
≥4	1	2	0.016	100	94.4	66.7	100	94.4
≥5	0	1	0.100	50	100	100	94.7	50.0
Relative indication score
≥1	7	2	0.189	100	61.1	22.2	100	61.1
≥2	2	1	0.284	50	88.9	33.3	94.1	44.5
Relative+absolute indication score
≥1	7	2	0.189	100	61.1	22.2	100	61.1
≥2	2	2	0.032	100	88.9	50	100	88.9
≥3	0	1	0.100	50	100	100	94.7	50.0

PPV, positive predictive value; NPV, negative predictive value; WF, worrisome feature; HRS, high-risk stigmata..

Since most studies investigating the natural history of CLPs are retrospective study performed without an elaborately defined protocol of surveillance,^4-14 knowledge about the long-term natural history of CLPs including the development of malignancy or surgical indication in each follow-up time is not sufficient yet. Our institutions have collected a cohort of CLPs with surveillance performed with a defined protocol. This study was performed using this cohort to investigate long-term clinical outcomes of CLPs. We documented critical clinical episodes such as progression to malignancy and surgical indication at each follow-up time to determine optimal duration and interval of surveillance for CLPs in clinical practice. To the best of our knowledge, most long-term surveillance in the current study were performed to provide data about long-term clinical outcomes of CLPs. Results of this study showed that malignancy progression could occur after 5 years of follow-up and that most CLPs without development of WF or HRS within 10 years had a favorable clinical course even after 10 years.

The optimal duration of surveillance has been a hot issue in the management of CLPs. Accumulating evidences suggest that CLPs might have a substantial risk of developing malignancy even after 5 years.^9,12,13 In the current study, while the risk of malignancy within 5 years was 0.4% (1 of 227 patients), the risk of malignancy after 5 years was 1.4% (3 of 210 patients). These results from previous and current studies provide evidence supporting continued surveillance after 5 years for patients with CLPs. However, evidence supporting surveillance after 10 years from initial diagnosis is not sufficient yet. In the current study, more than half of enrolled patients continued surveillance after 10 years. Among 109 patients without the development of WF or HRS at 10 years, no malignancy was developed. Although definite conclusion about the need or intensity of surveillance after 10 years cannot be made based on the outcome in the current study, most CLPs without development of WF or HRS within 10 years might have favorable clinical course even after 10 years. A further study investigating the need of surveillance for more than 10 years is warranted considering the cost-effectiveness of the surveillance, invasiveness of surgery, and patients’ comorbidities. We are continuing the surveillance for our cohort of patients enrolled in the current study. We plan to document and analyze clinical outcomes after 20 years of follow-up for our cohort.

The optimal interval of surveillance might also be an important issue in the management of CLPs. While few studies have addressed this issue,^6,19 a surveillance protocol including the optimal interval of follow-up has been suggested in a retrospective study using a cohort with IPMN in our institution.⁶ In that study, the authors suggested a follow-up interval from 6 months to 2 years based on initial size and growth rate of IPMN. Although it has been reported that growth rate is associated with the development of malignancy or surgical indication in previous and current studies,^11,20 accurate measurement of CLP size in millimeters using imaging studies might not always be easy in clinical practice, especially in a community hospital setting. Thus, knowledge of actual incidence of major clinical episodes at each follow-up time might provide more helpful information to determine the optimal interval of surveillance. In the current study, we investigated these episodes including malignancy, surgery, and surgical indication. We used surgical indication as a major indicator to determine the optimal interval of surveillance. This investigation showed that development to surgical indications did not occur within 12 months. It did not occur within 24 months in patients with initial CLP size less than 2 cm either. Results of the current study suggest that an interval of 12 months after the initial follow-up might be sufficient for CLPs with initial sizes of 2 to 3 cm and that an interval of 24 months after the initial follow-up might be sufficient for CLPs with initial sizes less than 2 cm (Fig. 5). However, as the number of enrolled patients in the current study was insufficient, further studies with larger series of CLPs addressing this issue are needed to validate our suggestions.

Figure 5. Suggested surveillance protocol throughout the follow-up based on the initial cyst size.
CLPs, cystic lesions of the pancreas; WF, worrisome feature; HRS, high-risk stigmata; CT, computed tomography; MRI, magnetic resonance imaging.

To date, several studies have validated the current consensus guidelines.^6,21-23 In the current study, validation was also performed for both the 2017 IAP guideline and the 2018 European guideline using our cohort of CLPs. Results of the current study showed that the diagnostic performance increased when risk factors were combined (WF score plus HRS score or relative indication plus absolute indication score) and that the diagnostic value of WF plus HRS scores was best among different risk scoring systems. Results of the current study suggest that the 2017 IAP guideline might be superior to the 2018 European guidelines. However, as numbers of patients who underwent surgery and those who experienced malignant progression in the current study were insufficient, further studies with larger series of CLPs addressing this issue are needed to validate our suggestion.

Compared to previous studies investigating the natural history of CLPs after long-term follow-up,^12,13 the incidence of malignancy in the current study appeared to be lower. This tendency of more benign clinical course in our study could be explained by several reasons. First, among our cohort of CLPs, we limited the inclusion criteria to CLPs without any WF or HRS at initial diagnosis. Second, while only CLPs with typical imaging findings of IPMN were enrolled in previous studies,^12,13 all CLPs without such typical IPMN findings were enrolled in the current study. Although advance has been made for high-resolution abdominal imaging and endoscopic ultrasonography,²⁴ an accurate differential diagnosis is difficult in some portions of CLPs in clinical practice.²⁵ This group of indeterminate CLPs includes premalignant CLPs such as IPMN and mucinous cystic neoplasm.²⁶ Thus, results of the current study using data including both typical IPMN and indeterminate CLPs might be more helpful for determining strategies of surveillance in real clinical practice.

The strengths and advantages of the current study are as follows. First, in the current study, we used our cohort, in which surveillance have been performed with a defined protocol. Using the data of the cohort, we could investigate major clinical episodes in each follow-up time. Second, this is the study providing data about most long-term clinical outcome of CLPs. More than half of enrolled patients continued surveillance after 10 years.

Despite its advantages, the current study has several limitations. First, our results have derived from a retrospective analysis and the definition of surgical indication might be somewhat arbitrary. We did not use this definition to identify group in which surgery should be needed. We used this definition to identify group in which caution of malignant progression should be needed during surveillance. Second, because we limited the inclusion criteria to patients who were diagnosed as CLPs from 2000 to 2008, the number of enrolled patients in the current study might not be sufficient. Thus, further studies with larger series of CLPs are needed to validate our results. Third, CLPs in our study might be heterogenous including both benign and premalignant CLPs. However, as mentioned earlier, results of our data including both typical IPMN and indeterminate CLPs might reflect real clinical practice and help determine strategies of surveillance in clinical practice. Finally, patients were followed up with CT scan every 6 months within the first 2 years in our study. Although, we used one-phase or low dose CT scan to minimize the radiation exposure, frequent CT radiation might cause harmful effect to patients. Most guidelines recommend magnetic resonance imaging as a follow-up modality to avoid the radiation hazard of CT.^15,17

In conclusion, while most CLPs had an indolent course during a long-term follow-up, the development of surgical indication and malignancy progression consistently occurred after 5 years of follow-up. No malignancy developed after 10 years of follow-up. Continued surveillance for CLPs after 5 years should be performed unless short life expectancy and significant comorbidities preclude surgery. Most CLPs without development of WF or HRS within 10 years might have favorable clinical course even after 10 years. For the first follow-up, an interval of 12 months might be sufficient for CLPs without WF or HRS and an interval of 24 months might be sufficient for CLPs with initial sizes less than 2 cm.

This study was supported by the Seoul National University Hospital Research Fund (Grant no. 3020220330).

No potential conflict of interest relevant to this article was reported.

Study concept and design: D.W.A., S.H.L., Y.T.K. Data acquisition: D.W.A., J.H.C., W.H.P. Data analysis and interpretation: D.W.A., J.H.C., I.R.C. Drafting of the manuscript: D.W.A. Critical revision of the manuscript for important intellectual content: S.H.L. Statistical analysis: D.W.A., D.K.J. Administrative, technical, or material support; study supervision: S.H.L., J.B.J., J.K.R., Y.T.K. Approval of final manuscript: all authors.

Supplementary materials can be accessed at https://doi.org/10.5009/gnl230017.