Follow-up of Incidentally Detected Pancreatic Cystic Neoplasms: Do Baseline MRI and CT Features Predict Cyst Growth?

Abstract

Background

Incidental detection of pancreatic cystic neoplasm (PCN) has increased. Since a small percentage of PCNs possess malignant potential, management is challenging. The recently revised American College of Radiology (ACR) recommendations define PCN measurement and growth for different categories based on baseline cyst size. However, no data are available regarding PCN growth rate under the ACR-defined size categories.

Purpose

To assess growth of incidentally detected PCNs on long-term imaging follow-up using revised ACR recommendations and to evaluate the association between baseline imaging features and growth.

Materials and Methods

This retrospective study included PCNs with baseline imaging performed between January 2002 and May 2017, with two or more cross-sectional imaging studies performed at least 12 months apart. PCN assessment was based on ACR 2017 recommendations. Cyst features, including location, septations, and mural nodules and multiplicity, were noted. Time to cyst progression (growth by ACR criteria) was examined by using baseline PCN size, among other factors.

Results

A total of 646 cysts in 390 patients were followed up for a median of 50 months (range, 12–186 months). A total of 184 (28.5%) cysts increased in size, 52 (8.1%) decreased in size, and 410 (63.4%) remained stable. For groups in which baseline PCN size was smaller than 5 mm, 5–14 mm, 15–25 mm, and larger than 25 mm, growth was noted in seven (13.2%), 106 (28.9%), 49 (32.2%), and 22 (29.7%) cysts, respectively. ACR baseline size categories (subhazard ratio: 2.8 [5–14-mm PCN group], 3.4 [15–25-mm PCN group], and 2.7 [>25 mm group], as compared with the <5 mm PCN group; P < .05 for each) demonstrated association with growth. Presence of mural nodules, septations, or lesion multiplicity failed to demonstrate association with growth. Among PCNs smaller than 5 mm at baseline, 100% of PCNs at 3-year follow-up and 94.2% of PCNs at 5-year follow-up were likely to remain stable.

Conclusion

American College of Radiology baseline size category of 15–25-mm pancreatic cystic neoplasms (PCNs) demonstrated the highest (3.1 times) likelihood of growth, as compared with the category of PCNs smaller than 5 mm. PCNs smaller than 5 mm at baseline did not demonstrate growth at 3-year imaging follow-up.

© RSNA, 2019

Online supplemental material is available for this article.

Summary

Pancreatic cystic neoplasms 15–25 mm in size were most likely to show growth on follow-up images obtained with CT or MRI; however, very small lesions (<5 mm) showed no interval growth until at least 3 years after initial lesion detection with CT or MRI.

Key Points

• ■ For pancreatic cystic neoplasm (PCN), American College of Radiology size category at baseline was an independent predictor of growth.

• ■ PCNs that were 15–25 mm at baseline showed the highest (3.1 times) likelihood of growth, as compared with those that were smaller than 5 mm at baseline.

• ■ Incidentally detected PCNs smaller than 5 mm at baseline did not demonstrate growth until at least 3 years of imaging follow-up, suggesting a 3-year follow-up examination may be adequate for very small PCNs.

• ■ After a median 50-month follow-up, we observed growth in 13% of PCNs smaller than 5 mm at baseline and approximately 30% of all cysts larger than 5 mm at baseline.

Introduction

With widespread use of cross-sectional imaging and ever-increasing advances in both CT and MRI techniques, the incidental detection of pancreatic cystic neoplasms (PCNs) has increased, with a reported prevalence of 2.2% for upper abdominal CT examinations and 19.6% for MRI examinations (1–4). Since a small percentage of PCNs progress to invasive cancers, the identification of PCNs is often a source of anxiety for patients and leads to challenging management decisions for involved physicians. The most commonly used guidelines (5,6) for management of intraductal papillary mucinous neoplasms place substantial emphasis on cyst size at baseline and on changes in size at follow-up using a combination of single-dimension size cut-offs, without standards for measurements or definition of what constitutes growth; most are consensus based rather than evidence based.

There are concerns that baseline size of a PCN may not be as important in predicting probability of malignancy present within the PCN at the time of evaluation as previously thought (7–10). Further, studies have suggested that the rate of cyst growth might be an important predictor of the likelihood that a PCN will progress to invasive cancer (11,12). Complicating this, there are no definitive suggestions regarding management approach in patients with multifocal PCNs. A recent prospective population-based cohort study revealed no pancreatic cancer during 5-year follow-up, while prospective follow-up imaging showed minimal progress in more than 50% of lesions.

The American College of Radiology (ACR) Incidental Findings Committee recently released a White Paper revising their recommendations on managing incidental pancreatic cysts (13). This White Paper outlined cyst measurement technique and recommended that a single measurement of the greatest length of the cyst in the long axis on either the axial image or the coronal image be recorded. In the new guidelines, growth at follow-up imaging is determined as follows: for cysts smaller than 5 mm, growth is represented by a 100% increase in long-axis diameter; for cysts 5–14 mm, growth is represented by a 50% increase in long-axis diameter; and for cysts 15 mm or larger, growth is represented by a 20% increase in long-axis diameter. In addition, the White Paper now recommends follow-up for 9–10 years in most patients. However, the optimal surveillance strategy for PCNs and the parameters to be monitored are not defined at this time. Moreover, data are not available regarding growth of PCNs as assessed using the latest White Paper recommendations.

Thus, the objective of this study was to assess the growth of incidentally detected pancreatic cystic neoplasms using definitions of growth defined in the latest American College of Radiology White Paper recommendations and to evaluate the association of PCN growth with baseline imaging features in long-term imaging follow-up.

Materials and Methods

Study Sample

This retrospective study was performed in compliance with the Health Insurance Portability and Accountability Act after we obtained approval from the institutional review board. Informed patient consent was waived.

We included patients with incidentally detected PCNs who underwent at least two abdominal MRI or CT studies performed at least 12 months apart. A retrospective review of images from routine MRI or CT on our picture archiving and communication system (Carestream Health, Rochester, NY) was performed. Consecutive asymptomatic patients older than 18 years with a clinical diagnosis of incidental PCN and without a strong family history of pancreatic cancer or a known predisposing germline mutation (14) or clinical suspicion for PCN were identified starting in January 2002. Patients with baseline imaging up to May 2017 were included to ensure a minimum follow-up of 12 months (until May 2018).

Imaging Technique

Imaging studies were performed with multidetector CT and MRI units from different vendors, including Siemens (Erlangen, Germany) and Philips (Amsterdam, the Netherlands), and two magnet strengths (1.5 and 3.0 T).

The MRI protocol included axial T2-weighted fat-saturated images (repetition time msec/echo time msec, 4000–6000/90–100; section thickness, 4–6 mm) and at least six thick slabs (4500/500–700; section thickness, 40–50 mm) of heavily T2-weighted MR cholangiopancreatography images obtained in the straight coronal plane. The sequence was repeated in the coronal oblique planes. Thin-section (echo time, 150–200 msec; section thickness, 1–3 mm; overlap, 0%–50%) respiratory-triggered images and navigator coronal three-dimensional fat-suppressed turbo spin-echo images also were obtained. Included studies varied with respect to presence of contrast-enhanced images and secretin-enhanced MR cholangiopancreatography images, as readers did not use these for image analysis.

The parameters used for unenhanced CT were 120 kVp, 250 mAs, 0.6-mm detector collimation, 0.75–1.0-mm section thickness, and 0.5-mm section interval. Images were then reconstructed at 5-mm thickness. These reconstructed images were then used for clinical reading. However, to more accurately measure the small size changes among the PCNs, readers were free to use the source thin-section images with a 0.75–1.0-mm thickness. Contrast-enhanced axial and coronal CT images were acquired with bolus triggering (arterial phase) and at 60 seconds (portal venous phase) after administration of intravenous contrast material by using bolus injection of 100–120 mL of iodinated contrast material (Visipaque 320 or Omnipaque 350; GE Healthcare, Waukesha Wis) using the same section and interval as those used for unenhanced CT.

Image Analysis

All images were reviewed, and measurements were performed by one of the three independent readers (P.P., A.P., and Y.L.) with 1, 4, and 5 years of experience in interpreting abdominal MRI, respectively. Prior to the study, all readers participated in an interactive training session with an experienced radiologist (I.R.K.) with 20 years of experience reading abdominal cross-sectional images, wherein standardized techniques to measure PCNs were outlined and agreed on. Readers then performed practice measurements on 10 cysts and were free to ask questions to resolve any conflicts. Diametric measurements were performed manually on a subjectively identified cross section with the largest diameter using outer wall–to–outer wall positioning of electronic calipers on axial and coronal images acquired with T2-weighted MRI or CT. The three readers then performed measurements on 50 randomly selected PCNs to assess interreader reproducibility for size measurements. To assess intrareader reproducibility, one reader repeated the measurements on the same group of 50 PCNs after 8 weeks to reduce the recall bias. The single largest diameter at baseline and at the last available imaging follow-up were recorded for all patients. For PCNs demonstrating size change, measurements were performed for all intermediate follow-up CT and MRI studies. In patients with multiple cysts, up to five of the largest cysts were identified and measured. Other cyst features, including location, septations, and mural nodules, were noted, and each cyst was given a cyst identification number. Medical records of all included cysts were reviewed, and histologic diagnosis for resected cysts was made available. Size change for each cyst was defined based on the recommendations of the latest White Paper of the ACR Incidental Findings Committee (13). For groups with baseline PCN size of less than 5 mm, 5–14 mm, 15–25 mm, or more than 25 mm, an increase or decrease from baseline size of 100% or more, 50% or more, 20% or more, and 20% or more, respectively, was considered a change in size. All the measurements in a given patient were performed by the same radiologist.

Statistical Analysis

To assess the interobserver variability of PCN size measurements in our study, intraclass correlation was calculated by using a two-way random-effects model for absolute agreement. To assess intrareader reproducibility, with the reader being a fixed effect in the model, the intraclass correlation coefficient (ICC) was calculated by using a two-way mixed-effects model for absolute agreement. The following scale was used to interpret the ICC: An ICC of less than 0.40 indicated poor agreement; an ICC of 0.40–0.59, fair agreement; an ICC of 0.60–0.74, good agreement; and an ICC of 0.75–1.00, excellent agreement (15). To assess agreement between the three readers for the ACR-defined baseline size categorization, κ statistic measurement of interrater agreement was used. Summary statistics were presented as mean ± standard deviation, range, median, interquartile range, or a percentage, as appropriate. Generalized estimating equations were used while accounting for the nesting of multiple PCNs within patients to model the PCN size change ([PCN size at last imaging–PCN size at baseline imaging]/follow-up duration) and PCN growth rate (PCN size change/follow-up duration). PCN size change and PCN growth rate were then modeled separately with PCN size at baseline categorized by using (a) ACR-defined categories and (b) increments of 1 mm as covariates. Generalized estimating equations were performed by using the xtgee command (Stata Software, College Station, Tex) with exchangeable correlation structure and Huber-White sandwich estimator of variance (with vce [robust] command) to allow valid standard errors even if correlations within the group were not as hypothesized by the specified correlation structure (16,17). Time to progression was examined by using baseline PCN size (ACR categories vs 1-mm increments), as well as other factors that have been variably reported in the literature to be associated with PCN growth (including presence of septations within PCNs [18], presence of a mural nodule within a PCN [19], and number of PCNs [1–2 vs ≥3]) (20) as predictors. Growth (according to ACR guidelines) was used as an outcome variable. Time without growth was calculated as the interval between baseline imaging and either the first subsequent imaging study that demonstrated growth by ACR definition (event, n = 184) or the last imaging examination performed before May 2018 (n = 441) or PCN surgery (n = 21) (both of which were set as censored in the absence of ACR-defined growth). The analysis was performed by using a competing-risks regression model (for the subhazard function of failure event of primary interest [ie, growth according to ACR criteria] while accounting for competing risk of patients undergoing PCN surgery that would completely preclude further cyst growth) while accounting for multiple observations within a patient. This was done by using the method of Fine and Gray (21). The covariates were initially explored as univariable predictors for any independent association with cyst growth. This was followed by multivariable analysis that included baseline PCN size (both as a continuous variable in 1-mm increments and by ACR categories, as assessed in separate models) and variables likely to be associated with the outcome of interest (cyst growth over time) as covariates. These potential confounders included presence of septations within the PCN, presence of a mural nodule within the PCN, number of PCNs (1–2 vs ≥3), PCN location, age, and sex. The multivariable models were assessed for the assumption of collinearity using tolerance and variance inflation factor. Two-sided P < .05 was considered to indicate a significant difference. Given the exploratory nature of the analysis, we chose not to control for false discovery rate; thus, we did not correct P values to control for type I errors. Statistical software (Stata, version 14; Stata Software) was used to perform all statistical analyses.

Results

Patient Demographics

A total of 390 patients with incidentally detected PCNs and without any known predisposing risk factor or clinical suspicion for PCN were included (Fig 1). The mean age of patients in the study cohort was 68 years ± 11, with 136 (35%) men and 254 (65%) women. The median follow-up duration for the cohort was 50 months (range, 12–186 months). The median number of cross-sectional imaging studies performed was four (interquartile range, three to six studies; range, two to 13 studies). Overall, 21 patients underwent surgical resection for PCN after a median follow-up of 31 months (range, 12–137 months). Pathologic analysis revealed one invasive adenocarcinoma arising within the PCN, 15 IPMNs with low-to-intermediate grade dysplasia, two IPMNs with high-grade dysplasia, two serous cystadenomas, and a mucinous cystic neoplasm without invasive carcinoma or high-grade dysplasia.

Figure 1:

Flowchart shows the included study sample with incidentally detected pancreatic cystic neoplasms (PCNs) with at least two abdominal MRI or CT examinations performed at least 12 months apart. CECT = contrast-enhanced CT, MRCP = MR cholangiopancreatography, T2WI = T2-weighted imaging.

PCN Characteristics at Baseline

The characteristics of PCNs at baseline are summarized in the Table. At baseline CT or MRI, 646 cysts had a median cyst size of 11 mm (interquartile range, 7–18 mm). A total of 148 cysts (23.2%) had one or more septations, and six cysts (1.0%) had a mural nodule. A total of 325 patients (83.3%) had one or two PCNs, while 65 patients (16.7%) had more than two PCNs. A total of 53 (8.2%), 367 (56.8%), 152 (23.5%), and 74 (11.5%) PCNs were noted for the ACR size categories of less than 5 mm, 5–14 mm, 15–25 mm, and greater than 25 mm, respectively. Location of the cyst was within the pancreatic head, body, or tail in 311 (48.3%), 182 (28.3%), and 151 (23.4%) patients, respectively. Among the PCNs demonstrating growth according to ACR 2017 criteria, the median baseline cyst size was 12 mm (interquartile range, 7–18 mm).

Demographic and Imaging Characteristics and their Association with Cyst Size Change in Patients with Incidentally Detected Pancreatic Cystic Neoplasms

Note.—Unless indicated otherwise, data are number of patients, with percentages in parentheses. ACR = American College of Radiology.

*For baseline pancreatic cystic neoplasm size groups of smaller than 5 mm, 5–14 mm, 15–25 mm, and larger than 25 mm, an increase or decrease from baseline size of 100% or more, 50% or more, 20% or more, and 20% or more, respectively, was considered indicative of a change in size, per ACR criteria.

^†Data in parentheses are the interquartile range.

^‡Data in parentheses are the range.

Inter- and Intraobserver PCN Size Reproducibility

The intraclass correlation coefficient for inter- and intraobserver agreement between PCN size measurements was 0.962 (95% confidence interval [CI]: 0.934, 0.978) and 0.969 (95% CI: 0.945, 0.982), respectively, which was suggestive of excellent reproducibility of size measurements. Substantial agreement was noted between the three readers for the ACR-defined baseline size categorization (κ statistic, 0.80; P < .001).

Factors Associated with PCN Growth

Please see Appendix E1 (online) for data on PCN growth and growth rate characteristics. A total of 184 cysts (28.5%) (over a median follow-up period of 59.5 months) increased in size, 52 (8.1%) (over a median follow-up period of 42.0 months) decreased in size, and 410 (63.4%) (over a median follow-up period of 48.0 months) remained stable. In the baseline size groups of less than 5 mm, 5–14 mm, 15–25 mm, and more than 25 mm, size increase was noted in seven (13.2%), 106 (28.9%), 49 (32.2%), and 22 (29.7%) cysts, respectively (Table). Figure 2 shows the proportion of PCNs showing size change in the baseline size categories as defined by the ACR White Paper recommendations. ACR baseline size categories demonstrated association with growth (based on the ACR 2017 definition) at univariable analysis (subhazard ratio [SHR]: 2.8 in the 5–14-mm group, 3.4 in the 15–25-mm group, and 2.7 in the >25 mm group as compared with the <5 mm group; P = .02) (Figs 3, 4).

Figure 2:

Bar chart shows the proportion of pancreatic cystic neoplasms with a size change in the baseline size categories, as defined by the American College of Radiology White Paper recommendations (13).

Figure 3a:

Images from axial fat-saturated T2-weighted MRI shows size stability in incidentally detected pancreatic cystic neoplasms (PCNs) smaller than 5 mm at baseline in a 53-year man. (a) Incidentally detected PCNs were noted within the pancreatic head and body and measured 4.9 and 4.6 mm in January 2013. This patient was seen at the multidisciplinary incidental pancreatic cyst clinic at our institute and underwent regular imaging follow-up. (b) MRI performed in August 2017, 55 months after a, shows no growth (by the American College of Radiology definition) in these PCNs (5.7 and 5.9 mm, respectively).

Figure 4a:

Images from axial fat-saturated T2-weighted MRI show growth in an incidentally detected pancreatic cystic neoplasm (PCN) measuring 15–25 mm at baseline in a 55-year woman. (a) PCN within the pancreatic tail measured 23.1 mm in December 2007. (b) In May 2014 (77 months later), this PCN had grown (34.9 mm).

Figure 3b:

Images from axial fat-saturated T2-weighted MRI shows size stability in incidentally detected pancreatic cystic neoplasms (PCNs) smaller than 5 mm at baseline in a 53-year man. (a) Incidentally detected PCNs were noted within the pancreatic head and body and measured 4.9 and 4.6 mm in January 2013. This patient was seen at the multidisciplinary incidental pancreatic cyst clinic at our institute and underwent regular imaging follow-up. (b) MRI performed in August 2017, 55 months after a, shows no growth (by the American College of Radiology definition) in these PCNs (5.7 and 5.9 mm, respectively).

Figure 4b:

Images from axial fat-saturated T2-weighted MRI show growth in an incidentally detected pancreatic cystic neoplasm (PCN) measuring 15–25 mm at baseline in a 55-year woman. (a) PCN within the pancreatic tail measured 23.1 mm in December 2007. (b) In May 2014 (77 months later), this PCN had grown (34.9 mm).

When we examined time to progression using baseline size at 1-mm increments as a continuous predictor and ACR guidelines for growth, we failed to find evidence of association of baseline size with growth based on ACR 2017 growth definitions (P = .46). Among demographic and other baseline cyst characteristics, while we failed to find evidence of association of age (P = .86), sex (P = .13), presence of a septation (P = .21), presence of more than two PCNs (P = .89), or cyst location (P = .29) with size change, presence of mural nodule (P = .07; SHR, 2.7) demonstrated tendency for association with size change. On multivariable analysis, ACR baseline size categories continued to demonstrate association with growth (SHR: 2.7 in the 5–14-mm group, 3.1 in the 15–25-mm group, and 2.3 in the >25 mm group; P < .05 for each), while the mural nodule demonstrated a tendency for association with growth (SHR, 2.8; P = .055). Tests to see if the data met the assumption of collinearity indicated that multicollinearity was not a concern (variance inflation factor <2 for each of the included covariates in the model). Figure 5 shows comparison of the cumulative incidence function curves for PCN growth among different ACR baseline size categories.

Figure 5:

Graph shows comparison of the cumulative incidence function curves for pancreatic cystic neoplasm (PCN) growth (by American College of Radiology [ACR] definition) among different ACR baseline size categories. For baseline PCN size groups smaller than 5 mm, 5–14 mm, 15–25 mm, and larger than 25 mm, an increase from baseline size of 100% or more, 50% or more, 20% or more, or 20% or more, respectively, was considered indicative of growth, per ACR criteria.

The estimated percentage of PCNs without growth at 3 and 5 years, respectively, among the ACR baseline size categories was as follows: 100% and 94.2% (95% CI: 78.6%, 98.5%) for PCNs smaller than 5 mm; 93.7% (95% CI: 90.4%, 95.9%) and 78.1% (95% CI: 72.3%, 82.8%) for 5–14-mm PCNs; 82.3% (95% CI: 74.8%, 87.8%) and 64.8% (95% CI: 54.3%, 73.6%) for 15–25-mm PCNs; and 86.0% (95% CI: 73.7%, 92.8%) and 63.4% (95% CI: 46.1%, 76.4%) for PCNs larger than 25 mm.

PCNs Associated with Malignancy

Of the 21 patients who underwent surgical resection of the PCN, one (33%) of three patients with invasive carcinoma or high-grade dysplasia met ACR growth criteria versus eight (53%) of 15 patients with low-grade dysplasia.

Discussion

In this study, we retrospectively assessed growth of pancreatic cystic neoplasms (PCNs) in a single-institution cohort of 646 incidentally detected PCNs in 390 patients and reported growth based on the American College of Radiology (ACR) 2017 White Paper definitions of growth. Our results showed that among the incidentally detected PCNs smaller than 5 mm at baseline, none grew within 3 years. The ACR baseline size category demonstrated association with growth, with PCNs in the 15–25-mm baseline size category demonstrating the highest (3.1 times) likelihood of growth as defined by the ACR, as compared with the category of PCNs smaller than 5 mm at baseline.

For PCNs smaller than 5 mm, the ACR 2017 White Paper recommends that one follow-up CT or MRI study at 2 years that demonstrates stability is sufficient to stop surveillance (13). Most recent studies focusing on size change and growth among incidental PCNs have excluded small (<5 mm) PCNs while assessing growth at follow-up (22–24). Recent studies have reported the presence of invasive carcinoma or high-grade dysplasia in PCNs (presumed to be IPMNs) as small as 4 mm (25). In our cohort, we observed growth (as defined by the ACR as doubling in the long-axis diameter) in 13.2% of PCNs smaller than 5 mm (over a median follow-up period of 65 months). Importantly, our findings showed no growth as defined by ACR criteria in PCNs smaller than 5 mm until at least 3-year follow-up. According to ACR definitions of cyst growth, approximately 30% of PCNs larger than 5 mm demonstrated interval growth in this cohort. Since several of these lesions (13 of 49 and 22 of 22 among the 15–25-mm and >25 mm baseline size categories, respectively) approached the 30-mm size that is considered a worrisome feature with regard to incidentally detected PCN management (5), it underscores the need for prolonged follow-up of these PCNs.

Recent studies have emphasized the growth rate of cysts at serial imaging. Kang et al reported that cysts with a growth rate of more than 2 mm per year had a higher risk of malignancy (11). Similarly, Kwong et al (12) concluded that branch duct IPMN growth rates of 2 mm or more per year should be considered worrisome features for branch duct IPMN, suggesting an increased risk of malignancy. In our cohort, PCNs demonstrating growth as defined by the ACR 2017 criteria revealed an overall average growth of 8.5 mm, with an average growth rate of 1.4 mm per year. Interestingly, only one PCN in our study cohort was ultimately found to harbor an invasive carcinoma over the follow-up duration (among the 21 cases that underwent surgical resection after follow-up of more than 12 months).

Our results failed to show a definite association between the presence of a mural nodule within a PCN and the likelihood of growth according to ACR 2017 criteria (SHR, 2.8; P = .057). The presence of a mural nodule within a PCN is a well-recognized worrisome feature (5), with authors suggesting conservative management of small cysts without its presence (26,27). Lack of association of a mural nodule with growth in our study cohort is likely due to the limited number of PCNs with mural nodules in our study since we excluded PCNs with high-risk stigmata that were surgically resected within 12 months of baseline imaging. Also, we failed to find evidence of an association between the presence of more than two PCNs and growth in our study. This finding is different from those reported in some previous studies (28,29). However, other authors have shown findings consistent with those of our study by showing no significant difference in the prevalence of PCNs with high-grade dysplasia or associated invasive carcinoma between the groups with solitary and multifocal PCNs (30) and concluded that multifocal branch duct IPMNs show slow growth and evolution over time (20). Nevertheless, our main focus was to investigate the association of PCN growth over time with baseline PCN size. Other associations presented in this article should be considered exploratory.

Our study had limitations. First, this was a single-center retrospective study. Second, although we studied the growth of PCNs on long-term follow-up and included the histologic findings from patients who underwent surgical resection of their PCNs, patient outcome and survival were not assessed. Third, the minimum imaging follow-up period for inclusion was 12 months; thus, size change assessment in some individuals was limited. Fourth, our study included PCNs with cross-modality follow-up (CT or MRI) that could theoretically lead to some perceived or measured size differences, given the different spatial resolutions of CT and MRI (19,31,32). Furthermore, we excluded patients with a familial predisposition of PCNs; thus, our conclusions cannot be extended to PCNs in patients with a known predisposition to having pancreatic cysts. Further studies are needed to define growth among PCNs in this population subgroup. Also, patients in our study sample received their diagnosis between 2002 and 2017, which might have introduced cohort effects. Moreover, the growth rates should be interpreted keeping in mind that censoring and competing risks could not be accounted for in the generalized estimating equation model, while heterogeneity in follow-up time might have affected interpretation of growth estimates, which are a function of follow-up duration. Here, we are explicitly assuming that absolute and relative growth estimates are not a function of follow-up time and thus are not biased by variable lengths of follow-up, surgery, death, or lost follow-up. Lastly, we used arbitrary thresholds for continuous data, including both baseline size (independent variable) and growth (outcome variable), which can reduce power, obscure estimation, and give the false impression that these arbitrary thresholds are meaningful (33,34). However, we only evaluated those thresholds put forth in recent ACR guidelines.

In conclusion, we report absolute growth and growth rate characteristics over long-term follow-up imaging of incidentally detected pancreatic cystic neoplasms (PCNs) and report growth among these PCNs based on the revised American College of Radiology (ACR) recommendations. A PCN in the 15–25-mm baseline size category demonstrated the highest (3.4 times) likelihood of growth, as compared with a PCN in the smaller than 5 mm category. PCNs smaller than 5 mm at baseline did not demonstrate growth for up to 3 years of imaging follow-up, and this could define the first follow-up time point among this subgroup of PCNs.