Dig Dis Sci (2015) 60:2800–2806 DOI 10.1007/s10620-015-3679-8
ORIGINAL ARTICLE
Rapid Growth Rates of Suspected Pancreatic Cyst Branch Duct Intraductal Papillary Mucinous Neoplasms Predict Malignancy Wilson T. Kwong1 • Robert D. Lawson2 • Gordon Hunt3 • Syed M. Fehmi1 • James A. Proudfoot4 • Ronghui Xu5 • Andrew Giap6 • Raymond S. Tang7 • Ingrid Gonzalez1 • Mary L. Krinsky1 • Thomas J. Savides1
Received: 30 November 2014 / Accepted: 17 April 2015 / Published online: 30 April 2015 Ó Springer Science+Business Media New York 2015
Abstract Background The majority of branch duct intraductal papillary mucinous neoplasms (BD-IPMNs) are recommended for surveillance imaging based on consensus guidelines. However, growth rates that should prompt concern for malignant transformation of BD-IPMN are unknown. Aims To determine whether BD-IPMN growth can predict an increased risk of malignancy and define growth rates concerning for malignant BD-IPMN. Methods The study is a retrospective, multicenter study of suspected BD-IPMN patients undergoing imaging surveillance. All patients underwent EUS evaluation followed by surveillance imaging. & Wilson T. Kwong [email protected] 1
Division of Gastroenterology, University of California San Diego Health Sciences, 9500 Gilman Drive (MC 0956), La Jolla, CA 92093-0956, USA
2
Department of Gastroenterology, Naval Medical Center, 34800 Bob Wilson Dr., San Diego, CA 92134, USA
3
Division of Gastroenterology, Kaiser Permanente San Diego Medical Center, 5893 Copley Dr., San Diego, CA 92111, USA
4
Biostatistics Unit, Clinical and Translational Research Institute, 9444 Medical Center Dr., La Jolla, CA 92037, USA
5
Department of Family and Preventive Medicine, and Mathematics, University of California, San Diego, 9500 Gilman Drive (MC 0112), La Jolla, CA 92093, USA
6
Division of Gastroenterology, Kaiser Permanente Anaheim Medical Center, Kraemer Medical Office 1, 3460 E. La Palma Ave, Anaheim, CA 92806, USA
7
Institute of Digestive Disease, The Chinese University of Hong Kong, 30-32 Ngan Shing St, Shatin, Hong Kong
123
Results Two hundred and eighty-four patients with suspected BD-IPMN without worrisome features or high-risk stigmata were followed for a median 56 months and underwent a median of four imaging studies. Nine patients (3.2 %) developed malignant BD-IPMN. Malignant BDIPMN grew at a faster rate (18.6 vs. 0.8 mm/year; P = 0.05) compared to benign BD-IPMN. BD-IPMN growth rate between 2 and 5 mm/year was associated with an increased risk of malignancy with hazard ratio (HR) of 11.4 (95 % CI 2.2–58.6) when compared to subjects with BD-IPMN growth rate \2 mm/year (P = 0.004). BDIPMN growth rate C5 mm/year had a hazard ratio of 19.5 (95 % CI 2.4–157.8) (P = 0.005). BD-IPMN growth rate of 2 mm/year had a sensitivity of 78 %, specificity of 90 %, and accuracy of 88 % to identify malignancy. Total BD-IPMN growth was also associated with increased risk of malignancy (P = 0.003) with all malignant IPMNs growing at least 10 mm prior to cancer diagnosis. Conclusions BD-IPMN growth rates C2 mm/year and total growth of C10 mm should be considered worrisome features for BD-IPMN at increased risk of malignancy. Keywords Branch duct intraductal papillary mucinous neoplasm Pancreatic cancer Pancreatic cyst Growth rate IPMN
Introduction Branch duct intraductal papillary mucinous neoplasms (BD-IPMNs) are pancreatic cysts that harbor malignant potential. Pooled data from previous studies estimate the frequency of malignancy of BD-IPMN to be 24 % (range 6.3–46.5 %) [1]. The optimal management of BD-IPMN remains controversial and continues to evolve with revised
Dig Dis Sci (2015) 60:2800–2806
management guidelines [1, 2]. Management of BD-IPMN has shifted away from surgical resection over the last decade, favoring a conservative approach with surveillance imaging for those without high-risk features. Revised international consensus guidelines [1] (Fukuoka guidelines) as well as recent guidelines from the American Gastroenterological Association (AGA) [2] recommend imaging surveillance for the majority of BD-IPMNs in the absence of worrisome features (thickened cyst walls, main pancreatic duct size greater than 5 mm, mural nodule, and abrupt change in caliber of pancreatic duct with distal pancreatic atrophy) or high-risk stigmata (enhancing solid component or pancreatic duct dilation greater than 10 mm). Although imaging surveillance is recommended for the majority of BD-IPMNs, the optimal surveillance strategy and which parameters should be monitored are uncertain at this time. There are no guideline recommendations regarding what amount or rate of BD-IPMN growth is worrisome for malignant transformation. During imaging surveillance of BD-IPMN, cyst size is often still the parameter that is monitored despite accumulating evidence that this is a poor predictor of malignancy [3–5]. Despite modest improvements in the ability to predict malignancy in BD-IPMN in the revised Fukuoka guidelines, more accurate predictors of malignancy are still needed. A surgical series reported a 34 % rate of malignancy among resected BD-IPMNs based on the Fukuoka guidelines, meaning that 66 % of patients underwent unnecessary resection [3]. As cyst size is the result of cyst growth over time, assessment of BD-IPMN growth rates may provide an earlier and more accurate indicator of malignant transformation compared to cyst size. A previous study [6] identified a higher 5-year risk of malignancy among BD-IPMN with growth rates greater than 2 mm per year (mm/year). Another study reported a 25 % rate of malignancy among BD-IPMN undergoing resection prompted by cyst growth during surveillance, which was comparable to a 23 % rate of malignancy in BD-IPMN resections due to consensus guideline criteria [7]. Monitoring of BD-IPMN growth rates may provide an additional method of identifying the malignant transformation of a BD-IPMN. The aim of this study was to determine whether BD-IPMN growth rates measured from serial cross-sectional imaging could accurately identify BDIPMN at increased risk of the development of malignancy.
Methods Study Population This is a multicenter, retrospective study of clinical and radiographic data from four medical centers in the same geographic region (University of California San Diego, San
2801
Diego Veterans Affairs medical center, Kaiser Permanente San Diego, and Kaiser Permanente Orange County). Patients were identified through a search of endoscopic databases for patients who underwent EUS evaluation of pancreatic cysts during a 10-year period from January 2002 to December 2011. Inclusion Criteria Patients diagnosed with suspected BD-IPMN cysts without Fukuoka guideline [1] high-risk stigmata (enhancing mural nodule, pancreatic duct C10 mm) or worrisome features (thickened cyst walls, main pancreatic duct [5 mm, non-enhancing mural nodule, or abrupt change in caliber of pancreatic duct with distal pancreatic atrophy) on initial cross-sectional imaging were included in this study. Patients with BD-IPMN larger than 3 cm were included if other worrisome features were absent as Fukuoka guidelines allow for surveillance of these BD-IPMNs. The presumptive diagnosis of BD-IPMN was based on morphology on CT and/or MRI in combination with EUS that was performed on all study patients. BD-IPMN was diagnosed based on the presence of unilocular or multilocular cysts of the pancreas and a non-dilated (B 5 mm) main pancreatic duct on EUS. As CT, MRI, and EUS reports do not consistently comment on communication of the cyst with the pancreatic duct, and ERCP was not done on all patients, diagnosis of BD-IPMN did not strictly require report of connection to the pancreatic duct. All patients were required to have at least one year of documented clinical follow-up to be included in the study. Patients were included if they had at least two cross-sectional imaging studies (CT, MRI, and EUS) at least 6 months apart. Exclusion Criteria Patients were excluded if they had any of the Fukuoka guideline ‘‘high risk’’ (jaundice, solid component, main pancreatic duct C10 mm) or ‘‘worrisome features’’ (thickened cyst wall, main duct size 5–9 mm, abrupt change in caliber of pancreatic duct). Patients with evidence of chronic pancreatitis on EUS/CT/MRI or a prior diagnosis of pancreatic cancer were excluded. Patients with multicystic ‘‘honeycomb’’-type cysts suspected to be serous cystadenomas were excluded. Patients with documented acute pancreatitis in the 6 months prior to cyst discovery were excluded to avoid inclusion of pseudocysts. Those undergoing pancreatic surgery for their cyst within the first 6 months of cyst discovery were excluded. Patients with a predominant mass with a smaller cystic component were also excluded.
123
2802
Dig Dis Sci (2015) 60:2800–2806
Data Collection and Analysis
Results
EUS examinations were performed by six experienced endosonographers who completed dedicated advanced endoscopy fellowships, and each had performed more than 1000 lifetime EUS examinations and practiced endosonography for greater than 5 years. Linear and radial echoendoscopes (Olympus Medical Systems, Center Valley, PA) were chosen at the discretion of the endoscopist. CT and MRI equipments utilized were not standardized as this study was performed across four medical centers over a 10-year span. In general, state-of-the-art equipment for the time period was utilized. Results from all available CT, MRI, MRCP, and EUS were examined for cyst size, main pancreatic duct size, cyst location, cyst morphology, presence of a mural nodule or solid component, cyst connection to main pancreatic duct, lymphadenopathy, thickened cyst walls, and cyst calcifications. Cyst aspiration was performed at the discretion of the endoscopist. BD-IPMN size measurements were recorded from all available cross-sectional imaging studies. Cyst size was recorded as the maximum cyst diameter on cross-sectional imaging. Chart review of a patient’s clinical course pertaining to the BD-IPMN was reviewed in detail. Symptoms possibly, but not definitively, attributable to the BD-IPMN were also recorded including abdominal pain, elevated liver function tests, jaundice, weight loss, pancreatitis, and nausea/vomiting. For BD-IPMN diagnosed with malignancy, only cyst size measurements prior to the malignancy diagnosis were included. Malignant BD-IPMN was defined as the presence of invasive carcinoma (based on surgical pathology or FNA cytology) arising from the pancreatic cyst. All other BD-IPMNs were categorized as benign.
Patient Characteristics
Statistical Analysis Cox proportional hazards regression model was used to quantify the effect of BD-IPMN growth rate and total growth on the time to cancer diagnosis, as well as the association between patient characteristics and time to cancer diagnosis. BD-IPMN growth rate was expressed first as a continuous time-varying covariate since baseline and then as a categorical time-varying covariate with three levels: \2 mm/year, between 2 and 5, and C5 mm/year. P values of \0.05 were considered statistically significant. The Kaplan–Meier estimate was used to derive the median duration of clinical follow-up, treating cancer diagnosis as censoring for follow-up. Data were analyzed using the R statistical computing environment (version 3.1.1) [8] and the R package ‘‘survival’’ [9].
123
Two hundred and eighty-four patients with suspected BDIPMN without high-risk stigmata or worrisome features met the study criteria and were included in the analysis. Clinical characteristics of the study population are shown in Table 1. Mean age of study patients was 67 ± 11 years. The mean age of patients with benign BD-IPMN was 67 years, while mean age of patients with malignant BDIPMN was 73 (Cox regression P = 0.07). Median duration of clinical follow-up was 56 [interquartile ratio (IQR) 36–74] months, and median duration of cross-sectional imaging surveillance was 36 (IQR 19–62) months. Each patient underwent a median of four cross-sectional imaging studies during surveillance. Three patients in this study underwent surgical resection (two pancreaticoduodenectomies and one distal pancreatectomy) with pathology demonstrating BD-IPMN with low-grade dysplasia in all three cases. Nine (3.2 %) patients developed invasive pancreatic adenocarcinoma arising from their BD-IPMN (Table 2). BD-IPMN Cyst Size, Growth, and Growth Rates Mean initial size of BD-IPMN cysts was 17.7 ± 10.7 mm, and mean final size was 19.3 ± 13.1 mm. Mean change in size was 1.6 ± 7.3 mm. Thirty (10.6 %) BD-IPMNs were initially C30 mm in size, and 37 (13.0 %) BD-IPMNs were C30 mm on final size. Mean initial size of benign BDIPMN was 17.4 ± 10.1 mm, while initial size of malignant BD-IPMN was 28.0 ± 22.0 mm (P = 0.19); mean final size of benign BD-IPMN was 18.3 ± 11.3 mm and for malignant BD-IPMN was 50.1 ± 23.4 mm (P = 0.004). Twenty-nine percent of suspected BD-IPMN increased in size, 57 % remained stable, and 14 % decreased in size (Fig. 1). An increase or decrease in size was defined as greater than 20 % increase or decrease from initial cyst size, respectively. Malignant BD-IPMN grew at a significantly faster rate compared to benign BD-IPMN (18.6 vs. 0.8 mm/year; P = 0.05). Malignant BD-IPMN grew by a significantly greater percentage (120 vs. 16 %; P = 0.009) and enlarged by a greater percentage per year compared to benign BD-IPMN (105 vs. 6 %/year; P = 0.06). Ten BD-IPMNs doubled in size (grew by 100 %) or more, and 5 (50 %) of these were malignant. Twenty-five BD-IPMNs grew at least 10 mm from initial detection and nine (36 %) were malignant. All nine malignant BD-IPMNs grew at least 10 mm since initial detection (Table 2). Three out of 5 (60 %) patients with a
Dig Dis Sci (2015) 60:2800–2806 Table 1 Patient demographics and presumptive BD-IPMN cyst characteristics
2803
All
Benign
Cancer
P value
N
284 (100 %)
275 (97 %)
9 (3 %)
Age (mean, year)
67.3 ± 10.8
67.1 ± 10.9
73.1 ± 4.9
Follow-up (median, month)
56
56
61
0.60
Male
123 (43 %)
120 (44 %)
3 (33 %)
0.56
Head
131 (46 %)
124 (45 %)
7 (78 %)
0.06
Body/tail
153 (54 %)
151 (55 %)
2 (22 %)
0.06
0.07
Location
BD-IPMN characteristics Initial size (mean, mm)
17.7 ± 10.7
17.4 ± 10.1
28.0 ± 22.0
0.19
Final size (mean, mm)
19.3 ± 13.1
18.3 ± 11.3
50.1 ± 23.4
0.004 0.17
30 (11 %)
27 (10 %)
3 (33 %)
Developed mural nodule
C3 cm initial size
15 (5 %)
11 (4 %)
4 (44 %)
0.05
Symptoms
68 (24 %)
63 (23 %)
5 (56 %)
0.90
Growth per year (mm/year)
Follow-up (months)b
Table 2 Characteristics of patients with malignant BD-IPMN Patient
Age (years)
Develop mural nodulea
Initial size (mm)
Final size (mm)
Total growth (mm)
Growth (%)
1
69
Yes
27
37
10
37
1.2
97
2
71
No
30
43
13
43
1.7
92
3 4
78 67
No Yes
80 11
90 24
10 13
13 118
2.6 2.9
46 57
5
70
Yes
11
40
29
264
5.6
62
6
76
No
40
63
23
58
14.5
19
7
75
Yes
12
26
14
117
24.0
7
8
70
No
14
46
32
229
54.9
7
9
82
No
27
82
55
204
60.0
13
a
Patients with mural nodules at initial EUS and CT imaging were excluded; these are patients who developed mural nodules during surveillance
b
Follow-up is the same as time to diagnosis of malignancy
Fig. 1 Serial size measurements of suspected benign and malignant BDIPMN
123
2804
Dig Dis Sci (2015) 60:2800–2806
mural nodule and growth rate C2 mm/year were found to be malignant. BD-IPMN Cyst Growth Rates as a Predictor of Invasive Carcinoma Using a Cox proportional hazard regression model with BD-IPMN growth rate as a time-dependent categorical variable, a BD-IPMN growth rate between 2 and 5 mm/ year is associated with an increased risk of invasive carcinoma diagnosis with a hazard ratio of 11.4 (95 % CI 2.2–58.6) when compared to BD-IPMN with growth rate \2 mm/year (P = 0.004). A BD-IPMN growth rate C5 mm/year is associated with an increased risk of invasive carcinoma diagnosis with hazard ratio of 19.5 (95 % CI 2.4–157.8) when compared to BD-IPMN with growth rate \2 mm/year (P = 0.005). Using BD-IPMN growth rate as a time-dependent continuous covariate, an increase in one unit (mm/year) of BD-IPMN growth rate is associated with an increased risk of invasive carcinoma diagnosis by a factor of 1.059 or 5.9 % (95 % CI 1.022–1.096; P = 0.001). The calculated area under ROC curve to assess the ability of BD-IPMN growth rate (mm/year) to predict malignancy is 0.94. The A 2 mm/year BD-IPMN growth rate predicted invasive carcinoma with sensitivity of 78 %, specificity of 90 %, and accuracy of 88 %. A 5 mm/year BD-IPMN growth rate predicted invasive carcinoma with sensitivity of 56 %, specificity of 97 %, and accuracy of 95 % (Table 3). Only two of 244 patients (0.8 %) with BD-IPMN growing slower than 2 mm/year had a malignant BD-IPMN. A Cox proportional hazard regression model with BDIPMN total growth (difference between the final and initial BD-IPMN size measurement) as a time-dependent continuous covariate showed an increase in one unit of cyst growth (1 mm) is associated with an increased risk of invasive carcinoma diagnosis by a factor of 1.201 or 20.1 % (95 % CI 1.064–1.356; P = 0.003). Using percentage growth of BD-IPMN (total growth divided by initial size)
Table 3 Diagnostic performance of BD-IPMN growth rates and total growth for predicting invasive carcinoma Sensitivity
Specificity
PPV
NPV
Accuracy
Growth rate 2 mm/year
78
90
18
99
88
5 mm/year
56
97
36
99
95
100
95
40
100
95
84
14
99
84
Total growth 10 mm
Percentage growth 40 %
78
PPV positive predictive value, NPV negative predictive value
123
to assess the presence of malignancy produces an area under ROC curve of 0.91. The optimal cutoff is 40 % BDIPMN growth, which produces a sensitivity, specificity, and accuracy of 78, 84, and 84 %, respectively.
Discussion International consensus (Fukuoka) guidelines published in 2012 recommend imaging surveillance of BD-IPMN without high-risk stigmata or worrisome features with the caveat that the correlation between rapid growth and malignancy is uncertain [1]. As most BD-IPMNs do not have mural nodules or pancreatic duct dilation, cyst size is often the parameter followed during surveillance despite several studies demonstrating that size is a poor predictor of malignancy [3–5]. Furthermore, the current consensus guidelines still do not exhibit sufficient accuracy as two-thirds of patients undergo resection for BD-IPMN without malignancy [3]. Therefore, there is a clear need for adjunctive, novel predictors of BD-IPMN malignancy to help improve the accuracy of future guidelines and minimize unnecessary surgeries. BD-IPMN growth and growth rates may be a more accurate predictor of malignancy that can be monitored by surveillance imaging. While no guideline recommendations currently exist regarding what amount of cyst growth is concerning for malignancy, most clinicians would be appropriately concerned regarding a rapidly enlarging BDIPMN. There are scant published data to guide interpretation of BD-IPMN growth rates. One study followed 58 patients with BD-IPMN using trans-abdominal ultrasound and reported an 11.3 mm/year growth rate detected 13 of 16 malignancies (carcinoma in situ and invasive carcinoma) with an 81 % accuracy [10]. A more recent study by Kang [6] examined 201 patients with low-risk BD-IPMN smaller than 30 mm with median follow-up of 28 months. The study reported a higher 5-year risk of malignancy for BD-IPMNs growing faster than 2 mm/year (44.1 vs. 0.8 %). The aim of this study was to assess whether growth rates of suspected BD-IPMN could identify BD-IPMNs at higher risk of developing malignancy. Our study population consisted of 284 patients with presumptive BD-IPMN without high risk or worrisome features that would be recommended for imaging surveillance consistent with the international consensus guidelines [1] and AGA guidelines [2]. We found a significant association between BD-IPMN growth rate and development of invasive carcinoma. A BD-IPMN growth rate between 2 and 5 mm/year is associated with an increased risk of malignancy with a hazard ratio of 11.4 compared to BD-IPMN growth rates less than 2 mm/year. BD-IPMNs with a growth rate C5 mm/year
Dig Dis Sci (2015) 60:2800–2806
have an increased risk of malignancy with a higher hazard ratio of 19.5 compared to BD-IPMN growth rate less than 2 mm/year. Based on our results, BD-IPMN total growth and growth rates appear promising in their ability to identify BD-IPMN at high risk of malignancy. However, there are limitations that should be noted. From a pragmatic standpoint, the ability to measure a 5-mm difference on imaging is likely to be more reliable than measuring a 2-mm difference. A prior study demonstrated that variability in pancreatic cyst size measurements between CT, MRI, and EUS studies is 3 to 4 mm [11]. Attempts to detect a 2-mm difference on imaging are more vulnerable to human errors in measurement, measurement of the cyst at different levels, and differences in imaging modalities. A 5 mm/year growth would be more reliably detected on imaging, has a higher hazard ratio for malignancy, and higher specificity for malignant transformation. A higher specificity is desirable when evaluating BD-IPMN for malignancy to avoid unnecessary resections. A 2 mm/year growth rate over two or more years would produce at least a 4-mm difference, which could be more reliably detected. Both 2 and 5 mm/ year cutoffs demonstrate a high negative predictive value of 99 % that could be utilized to reassure patients with slow growing BD-IPMNs regarding the benign nature of their cyst and also reduce the number of unnecessary resections. Another limitation of this study is the retrospective nature. Surveillance was not standardized as this was a retrospective study across four centers. Follow-up imaging was generally recommended on at least a yearly basis, though patients and referring physicians did not necessarily adhere to recommendations. The study also based the diagnosis of BD-IPMN on clinical and imaging characteristics, rather than surgical resection. However, surgical resection would have eliminated our ability to follow the natural history and growth of these cysts. The diagnosis of BD-IPMN based on clinical and EUS evaluation represents actual clinical practice. Routine cyst fluid aspiration for CEA and serum measurement of CA 19–9 was not performed in all cases as guidelines do not recommend these as part of routine evaluation. There was also potential selection bias as our study population was referred for EUS evaluation. However, despite a bias toward a higher-risk population, the overall rate of BD-IPMN malignancy was 3 %, which is consistent with prior studies [6] that also excluded worrisome and high-risk BD-IPMN. The patients in the study were followed clinically for a median 56 months and with imaging surveillance for 36 months. The imaging surveillance was shorter due to several reasons including the discontinuation of imaging surveillance due to stability of the cyst or the patient was no longer deemed a surgical candidate (due to age or
2805
comorbidities), or personal preference to discontinue surveillance. Overall, the length of both clinical and imaging surveillance is a strength of the study, allowing for the capture of the long-term natural history and malignant transformation of BD-IPMN. Resection is likely not necessary in all cases of BDIPMN enlarging faster than 2 mm/year, but these patients should be evaluated with EUS and receive closer surveillance. The 2 and 5 mm yearly growth rates can be followed over the course of multiple years (as opposed to just 1 year) as was done in this study which may produce a more reliable trend and indicator of malignancy. Should the cyst continue to exhibit significant growth or develop other worrisome features, then resection should be considered. We also found that a high percentage of patients with BD-IPMN that doubled in size or grew more than 10 mm since initial detection have a high rate of malignancy (50 and 36 %, respectively). A 40 % increase in cyst size has a sensitivity, specificity, and accuracy of 78, 84, and 84 %, respectively, for the detection of malignancy. The 2 mm/year growth rate and 40 % percentage growth both perform favorably in their ability to identify malignant BD-IPMN (Table 3) compared to the current consensus guidelines, which demonstrate a reported sensitivity, specificity, and accuracy of 72, 78, and 80 %, respectively [3]. In conclusion, in this multicenter study of patients with suspected BD-IPMN undergoing long-term surveillance imaging, we demonstrate that BD-IPMN growth rates greater than 2 and 5 mm/year are associated with an increased risk of malignant transformation. In addition, total BD-IPMN growth was also associated with an increased risk of malignancy with all malignant BD-IPMNs growing at least 10 mm prior to a cancer diagnosis. Suspected BDIPMN growth rates of 2 or 5 mm/year or a total growth of 10 mm should be considered for inclusion in future guidelines as worrisome features for malignant transformation of BD-IPMN. Perhaps of equal or greater value is the ability to reassure patients that BD-IPMN growing less than 2 mm/year have less than 1 % rate of malignancy during long-term surveillance and minimize unnecessary resections. Acknowledgments The study was partially supported by the National Institutes of Health Grant UL1TR000100 (Ronghui Xu). Conflict of interest
None.
References 1. Tanaka M, Ferna´ndez-del Castillo C, Adsay V, et al. International consensus guidelines 2012 for the management of IPMN and MCN of the pancreas. Pancreatology. 2012;12:183–197.
123
2806 2. Vege S, Ziring B, Jain R, et al. American gastroenterological association institute guideline on the diagnosis and management of asymptomatic neoplastic pancreatic cysts. Gastroenterology. 2015;148:819–822. 3. Jang JY, Park T, Lee S. Validation of international consensus guidelines for the resection of branch duct-type intraductal papillary mucinous neoplasms. Br J Surg. 2014;101:686–692. 4. Kim KW, Park SH, Pyo J, et al. Imaging features to distinguish malignant and benign branch-duct type intraductal papillary mucinous neoplasms of the pancreas; a meta-analysis. Ann Surg. 2014;259:72–81. 5. Shimizu Y, Yamaue H, Maguchi H, et al. Predictors of malignancy in intraductal papillary mucinous neoplasm of the pancreas: analysis of 310 pancreatic resection patients at multiple high-volume centers. Pancreas. 2013;42:883–888. 6. Kang MJ, Jang JY, Kim SJ, et al. Cyst growth rate predicts malignancy in patients with branch duct intraductal papillary mucinous neoplasms. Clin Gastroenterol Hepatol. 2011;9:87–93.
123
Dig Dis Sci (2015) 60:2800–2806 7. Bae SY, Lee KT, Lee JH, Lee JK, Lee KH, Rhee JC. Proper management and follow-up strategy of branch duct intraductal papillary mucinous neoplasms of the pancreas. Dig Liver Dis. 2012;44:257–260. 8. R Core Team. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2014. 9. Therneau T (2014) A package for survival analysis in S. R package version 2.37-7. 10. Yamaguchi T, Baba T, Ishihara T, et al. Long-term follow-up of intraductal papillary mucinous neoplasm of the pancreas with ultrasonography. Clin Gastroenterol Hepatol. 2005;3:1136–1143. 11. Maimone S, Agrawal D, Pollack MJ, et al. Variability in measurements of pancreatic cyst size among EUS, CT, and magnetic resonance imaging modalities. Gastrointest Endosc. 2010;71: 945–950.
ORIGINAL ARTICLE
Rapid Growth Rates of Suspected Pancreatic Cyst Branch Duct Intraductal Papillary Mucinous Neoplasms Predict Malignancy Wilson T. Kwong1 • Robert D. Lawson2 • Gordon Hunt3 • Syed M. Fehmi1 • James A. Proudfoot4 • Ronghui Xu5 • Andrew Giap6 • Raymond S. Tang7 • Ingrid Gonzalez1 • Mary L. Krinsky1 • Thomas J. Savides1
Received: 30 November 2014 / Accepted: 17 April 2015 / Published online: 30 April 2015 Ó Springer Science+Business Media New York 2015
Abstract Background The majority of branch duct intraductal papillary mucinous neoplasms (BD-IPMNs) are recommended for surveillance imaging based on consensus guidelines. However, growth rates that should prompt concern for malignant transformation of BD-IPMN are unknown. Aims To determine whether BD-IPMN growth can predict an increased risk of malignancy and define growth rates concerning for malignant BD-IPMN. Methods The study is a retrospective, multicenter study of suspected BD-IPMN patients undergoing imaging surveillance. All patients underwent EUS evaluation followed by surveillance imaging. & Wilson T. Kwong [email protected] 1
Division of Gastroenterology, University of California San Diego Health Sciences, 9500 Gilman Drive (MC 0956), La Jolla, CA 92093-0956, USA
2
Department of Gastroenterology, Naval Medical Center, 34800 Bob Wilson Dr., San Diego, CA 92134, USA
3
Division of Gastroenterology, Kaiser Permanente San Diego Medical Center, 5893 Copley Dr., San Diego, CA 92111, USA
4
Biostatistics Unit, Clinical and Translational Research Institute, 9444 Medical Center Dr., La Jolla, CA 92037, USA
5
Department of Family and Preventive Medicine, and Mathematics, University of California, San Diego, 9500 Gilman Drive (MC 0112), La Jolla, CA 92093, USA
6
Division of Gastroenterology, Kaiser Permanente Anaheim Medical Center, Kraemer Medical Office 1, 3460 E. La Palma Ave, Anaheim, CA 92806, USA
7
Institute of Digestive Disease, The Chinese University of Hong Kong, 30-32 Ngan Shing St, Shatin, Hong Kong
123
Results Two hundred and eighty-four patients with suspected BD-IPMN without worrisome features or high-risk stigmata were followed for a median 56 months and underwent a median of four imaging studies. Nine patients (3.2 %) developed malignant BD-IPMN. Malignant BDIPMN grew at a faster rate (18.6 vs. 0.8 mm/year; P = 0.05) compared to benign BD-IPMN. BD-IPMN growth rate between 2 and 5 mm/year was associated with an increased risk of malignancy with hazard ratio (HR) of 11.4 (95 % CI 2.2–58.6) when compared to subjects with BD-IPMN growth rate \2 mm/year (P = 0.004). BDIPMN growth rate C5 mm/year had a hazard ratio of 19.5 (95 % CI 2.4–157.8) (P = 0.005). BD-IPMN growth rate of 2 mm/year had a sensitivity of 78 %, specificity of 90 %, and accuracy of 88 % to identify malignancy. Total BD-IPMN growth was also associated with increased risk of malignancy (P = 0.003) with all malignant IPMNs growing at least 10 mm prior to cancer diagnosis. Conclusions BD-IPMN growth rates C2 mm/year and total growth of C10 mm should be considered worrisome features for BD-IPMN at increased risk of malignancy. Keywords Branch duct intraductal papillary mucinous neoplasm Pancreatic cancer Pancreatic cyst Growth rate IPMN
Introduction Branch duct intraductal papillary mucinous neoplasms (BD-IPMNs) are pancreatic cysts that harbor malignant potential. Pooled data from previous studies estimate the frequency of malignancy of BD-IPMN to be 24 % (range 6.3–46.5 %) [1]. The optimal management of BD-IPMN remains controversial and continues to evolve with revised
Dig Dis Sci (2015) 60:2800–2806
management guidelines [1, 2]. Management of BD-IPMN has shifted away from surgical resection over the last decade, favoring a conservative approach with surveillance imaging for those without high-risk features. Revised international consensus guidelines [1] (Fukuoka guidelines) as well as recent guidelines from the American Gastroenterological Association (AGA) [2] recommend imaging surveillance for the majority of BD-IPMNs in the absence of worrisome features (thickened cyst walls, main pancreatic duct size greater than 5 mm, mural nodule, and abrupt change in caliber of pancreatic duct with distal pancreatic atrophy) or high-risk stigmata (enhancing solid component or pancreatic duct dilation greater than 10 mm). Although imaging surveillance is recommended for the majority of BD-IPMNs, the optimal surveillance strategy and which parameters should be monitored are uncertain at this time. There are no guideline recommendations regarding what amount or rate of BD-IPMN growth is worrisome for malignant transformation. During imaging surveillance of BD-IPMN, cyst size is often still the parameter that is monitored despite accumulating evidence that this is a poor predictor of malignancy [3–5]. Despite modest improvements in the ability to predict malignancy in BD-IPMN in the revised Fukuoka guidelines, more accurate predictors of malignancy are still needed. A surgical series reported a 34 % rate of malignancy among resected BD-IPMNs based on the Fukuoka guidelines, meaning that 66 % of patients underwent unnecessary resection [3]. As cyst size is the result of cyst growth over time, assessment of BD-IPMN growth rates may provide an earlier and more accurate indicator of malignant transformation compared to cyst size. A previous study [6] identified a higher 5-year risk of malignancy among BD-IPMN with growth rates greater than 2 mm per year (mm/year). Another study reported a 25 % rate of malignancy among BD-IPMN undergoing resection prompted by cyst growth during surveillance, which was comparable to a 23 % rate of malignancy in BD-IPMN resections due to consensus guideline criteria [7]. Monitoring of BD-IPMN growth rates may provide an additional method of identifying the malignant transformation of a BD-IPMN. The aim of this study was to determine whether BD-IPMN growth rates measured from serial cross-sectional imaging could accurately identify BDIPMN at increased risk of the development of malignancy.
Methods Study Population This is a multicenter, retrospective study of clinical and radiographic data from four medical centers in the same geographic region (University of California San Diego, San
2801
Diego Veterans Affairs medical center, Kaiser Permanente San Diego, and Kaiser Permanente Orange County). Patients were identified through a search of endoscopic databases for patients who underwent EUS evaluation of pancreatic cysts during a 10-year period from January 2002 to December 2011. Inclusion Criteria Patients diagnosed with suspected BD-IPMN cysts without Fukuoka guideline [1] high-risk stigmata (enhancing mural nodule, pancreatic duct C10 mm) or worrisome features (thickened cyst walls, main pancreatic duct [5 mm, non-enhancing mural nodule, or abrupt change in caliber of pancreatic duct with distal pancreatic atrophy) on initial cross-sectional imaging were included in this study. Patients with BD-IPMN larger than 3 cm were included if other worrisome features were absent as Fukuoka guidelines allow for surveillance of these BD-IPMNs. The presumptive diagnosis of BD-IPMN was based on morphology on CT and/or MRI in combination with EUS that was performed on all study patients. BD-IPMN was diagnosed based on the presence of unilocular or multilocular cysts of the pancreas and a non-dilated (B 5 mm) main pancreatic duct on EUS. As CT, MRI, and EUS reports do not consistently comment on communication of the cyst with the pancreatic duct, and ERCP was not done on all patients, diagnosis of BD-IPMN did not strictly require report of connection to the pancreatic duct. All patients were required to have at least one year of documented clinical follow-up to be included in the study. Patients were included if they had at least two cross-sectional imaging studies (CT, MRI, and EUS) at least 6 months apart. Exclusion Criteria Patients were excluded if they had any of the Fukuoka guideline ‘‘high risk’’ (jaundice, solid component, main pancreatic duct C10 mm) or ‘‘worrisome features’’ (thickened cyst wall, main duct size 5–9 mm, abrupt change in caliber of pancreatic duct). Patients with evidence of chronic pancreatitis on EUS/CT/MRI or a prior diagnosis of pancreatic cancer were excluded. Patients with multicystic ‘‘honeycomb’’-type cysts suspected to be serous cystadenomas were excluded. Patients with documented acute pancreatitis in the 6 months prior to cyst discovery were excluded to avoid inclusion of pseudocysts. Those undergoing pancreatic surgery for their cyst within the first 6 months of cyst discovery were excluded. Patients with a predominant mass with a smaller cystic component were also excluded.
123
2802
Dig Dis Sci (2015) 60:2800–2806
Data Collection and Analysis
Results
EUS examinations were performed by six experienced endosonographers who completed dedicated advanced endoscopy fellowships, and each had performed more than 1000 lifetime EUS examinations and practiced endosonography for greater than 5 years. Linear and radial echoendoscopes (Olympus Medical Systems, Center Valley, PA) were chosen at the discretion of the endoscopist. CT and MRI equipments utilized were not standardized as this study was performed across four medical centers over a 10-year span. In general, state-of-the-art equipment for the time period was utilized. Results from all available CT, MRI, MRCP, and EUS were examined for cyst size, main pancreatic duct size, cyst location, cyst morphology, presence of a mural nodule or solid component, cyst connection to main pancreatic duct, lymphadenopathy, thickened cyst walls, and cyst calcifications. Cyst aspiration was performed at the discretion of the endoscopist. BD-IPMN size measurements were recorded from all available cross-sectional imaging studies. Cyst size was recorded as the maximum cyst diameter on cross-sectional imaging. Chart review of a patient’s clinical course pertaining to the BD-IPMN was reviewed in detail. Symptoms possibly, but not definitively, attributable to the BD-IPMN were also recorded including abdominal pain, elevated liver function tests, jaundice, weight loss, pancreatitis, and nausea/vomiting. For BD-IPMN diagnosed with malignancy, only cyst size measurements prior to the malignancy diagnosis were included. Malignant BD-IPMN was defined as the presence of invasive carcinoma (based on surgical pathology or FNA cytology) arising from the pancreatic cyst. All other BD-IPMNs were categorized as benign.
Patient Characteristics
Statistical Analysis Cox proportional hazards regression model was used to quantify the effect of BD-IPMN growth rate and total growth on the time to cancer diagnosis, as well as the association between patient characteristics and time to cancer diagnosis. BD-IPMN growth rate was expressed first as a continuous time-varying covariate since baseline and then as a categorical time-varying covariate with three levels: \2 mm/year, between 2 and 5, and C5 mm/year. P values of \0.05 were considered statistically significant. The Kaplan–Meier estimate was used to derive the median duration of clinical follow-up, treating cancer diagnosis as censoring for follow-up. Data were analyzed using the R statistical computing environment (version 3.1.1) [8] and the R package ‘‘survival’’ [9].
123
Two hundred and eighty-four patients with suspected BDIPMN without high-risk stigmata or worrisome features met the study criteria and were included in the analysis. Clinical characteristics of the study population are shown in Table 1. Mean age of study patients was 67 ± 11 years. The mean age of patients with benign BD-IPMN was 67 years, while mean age of patients with malignant BDIPMN was 73 (Cox regression P = 0.07). Median duration of clinical follow-up was 56 [interquartile ratio (IQR) 36–74] months, and median duration of cross-sectional imaging surveillance was 36 (IQR 19–62) months. Each patient underwent a median of four cross-sectional imaging studies during surveillance. Three patients in this study underwent surgical resection (two pancreaticoduodenectomies and one distal pancreatectomy) with pathology demonstrating BD-IPMN with low-grade dysplasia in all three cases. Nine (3.2 %) patients developed invasive pancreatic adenocarcinoma arising from their BD-IPMN (Table 2). BD-IPMN Cyst Size, Growth, and Growth Rates Mean initial size of BD-IPMN cysts was 17.7 ± 10.7 mm, and mean final size was 19.3 ± 13.1 mm. Mean change in size was 1.6 ± 7.3 mm. Thirty (10.6 %) BD-IPMNs were initially C30 mm in size, and 37 (13.0 %) BD-IPMNs were C30 mm on final size. Mean initial size of benign BDIPMN was 17.4 ± 10.1 mm, while initial size of malignant BD-IPMN was 28.0 ± 22.0 mm (P = 0.19); mean final size of benign BD-IPMN was 18.3 ± 11.3 mm and for malignant BD-IPMN was 50.1 ± 23.4 mm (P = 0.004). Twenty-nine percent of suspected BD-IPMN increased in size, 57 % remained stable, and 14 % decreased in size (Fig. 1). An increase or decrease in size was defined as greater than 20 % increase or decrease from initial cyst size, respectively. Malignant BD-IPMN grew at a significantly faster rate compared to benign BD-IPMN (18.6 vs. 0.8 mm/year; P = 0.05). Malignant BD-IPMN grew by a significantly greater percentage (120 vs. 16 %; P = 0.009) and enlarged by a greater percentage per year compared to benign BD-IPMN (105 vs. 6 %/year; P = 0.06). Ten BD-IPMNs doubled in size (grew by 100 %) or more, and 5 (50 %) of these were malignant. Twenty-five BD-IPMNs grew at least 10 mm from initial detection and nine (36 %) were malignant. All nine malignant BD-IPMNs grew at least 10 mm since initial detection (Table 2). Three out of 5 (60 %) patients with a
Dig Dis Sci (2015) 60:2800–2806 Table 1 Patient demographics and presumptive BD-IPMN cyst characteristics
2803
All
Benign
Cancer
P value
N
284 (100 %)
275 (97 %)
9 (3 %)
Age (mean, year)
67.3 ± 10.8
67.1 ± 10.9
73.1 ± 4.9
Follow-up (median, month)
56
56
61
0.60
Male
123 (43 %)
120 (44 %)
3 (33 %)
0.56
Head
131 (46 %)
124 (45 %)
7 (78 %)
0.06
Body/tail
153 (54 %)
151 (55 %)
2 (22 %)
0.06
0.07
Location
BD-IPMN characteristics Initial size (mean, mm)
17.7 ± 10.7
17.4 ± 10.1
28.0 ± 22.0
0.19
Final size (mean, mm)
19.3 ± 13.1
18.3 ± 11.3
50.1 ± 23.4
0.004 0.17
30 (11 %)
27 (10 %)
3 (33 %)
Developed mural nodule
C3 cm initial size
15 (5 %)
11 (4 %)
4 (44 %)
0.05
Symptoms
68 (24 %)
63 (23 %)
5 (56 %)
0.90
Growth per year (mm/year)
Follow-up (months)b
Table 2 Characteristics of patients with malignant BD-IPMN Patient
Age (years)
Develop mural nodulea
Initial size (mm)
Final size (mm)
Total growth (mm)
Growth (%)
1
69
Yes
27
37
10
37
1.2
97
2
71
No
30
43
13
43
1.7
92
3 4
78 67
No Yes
80 11
90 24
10 13
13 118
2.6 2.9
46 57
5
70
Yes
11
40
29
264
5.6
62
6
76
No
40
63
23
58
14.5
19
7
75
Yes
12
26
14
117
24.0
7
8
70
No
14
46
32
229
54.9
7
9
82
No
27
82
55
204
60.0
13
a
Patients with mural nodules at initial EUS and CT imaging were excluded; these are patients who developed mural nodules during surveillance
b
Follow-up is the same as time to diagnosis of malignancy
Fig. 1 Serial size measurements of suspected benign and malignant BDIPMN
123
2804
Dig Dis Sci (2015) 60:2800–2806
mural nodule and growth rate C2 mm/year were found to be malignant. BD-IPMN Cyst Growth Rates as a Predictor of Invasive Carcinoma Using a Cox proportional hazard regression model with BD-IPMN growth rate as a time-dependent categorical variable, a BD-IPMN growth rate between 2 and 5 mm/ year is associated with an increased risk of invasive carcinoma diagnosis with a hazard ratio of 11.4 (95 % CI 2.2–58.6) when compared to BD-IPMN with growth rate \2 mm/year (P = 0.004). A BD-IPMN growth rate C5 mm/year is associated with an increased risk of invasive carcinoma diagnosis with hazard ratio of 19.5 (95 % CI 2.4–157.8) when compared to BD-IPMN with growth rate \2 mm/year (P = 0.005). Using BD-IPMN growth rate as a time-dependent continuous covariate, an increase in one unit (mm/year) of BD-IPMN growth rate is associated with an increased risk of invasive carcinoma diagnosis by a factor of 1.059 or 5.9 % (95 % CI 1.022–1.096; P = 0.001). The calculated area under ROC curve to assess the ability of BD-IPMN growth rate (mm/year) to predict malignancy is 0.94. The A 2 mm/year BD-IPMN growth rate predicted invasive carcinoma with sensitivity of 78 %, specificity of 90 %, and accuracy of 88 %. A 5 mm/year BD-IPMN growth rate predicted invasive carcinoma with sensitivity of 56 %, specificity of 97 %, and accuracy of 95 % (Table 3). Only two of 244 patients (0.8 %) with BD-IPMN growing slower than 2 mm/year had a malignant BD-IPMN. A Cox proportional hazard regression model with BDIPMN total growth (difference between the final and initial BD-IPMN size measurement) as a time-dependent continuous covariate showed an increase in one unit of cyst growth (1 mm) is associated with an increased risk of invasive carcinoma diagnosis by a factor of 1.201 or 20.1 % (95 % CI 1.064–1.356; P = 0.003). Using percentage growth of BD-IPMN (total growth divided by initial size)
Table 3 Diagnostic performance of BD-IPMN growth rates and total growth for predicting invasive carcinoma Sensitivity
Specificity
PPV
NPV
Accuracy
Growth rate 2 mm/year
78
90
18
99
88
5 mm/year
56
97
36
99
95
100
95
40
100
95
84
14
99
84
Total growth 10 mm
Percentage growth 40 %
78
PPV positive predictive value, NPV negative predictive value
123
to assess the presence of malignancy produces an area under ROC curve of 0.91. The optimal cutoff is 40 % BDIPMN growth, which produces a sensitivity, specificity, and accuracy of 78, 84, and 84 %, respectively.
Discussion International consensus (Fukuoka) guidelines published in 2012 recommend imaging surveillance of BD-IPMN without high-risk stigmata or worrisome features with the caveat that the correlation between rapid growth and malignancy is uncertain [1]. As most BD-IPMNs do not have mural nodules or pancreatic duct dilation, cyst size is often the parameter followed during surveillance despite several studies demonstrating that size is a poor predictor of malignancy [3–5]. Furthermore, the current consensus guidelines still do not exhibit sufficient accuracy as two-thirds of patients undergo resection for BD-IPMN without malignancy [3]. Therefore, there is a clear need for adjunctive, novel predictors of BD-IPMN malignancy to help improve the accuracy of future guidelines and minimize unnecessary surgeries. BD-IPMN growth and growth rates may be a more accurate predictor of malignancy that can be monitored by surveillance imaging. While no guideline recommendations currently exist regarding what amount of cyst growth is concerning for malignancy, most clinicians would be appropriately concerned regarding a rapidly enlarging BDIPMN. There are scant published data to guide interpretation of BD-IPMN growth rates. One study followed 58 patients with BD-IPMN using trans-abdominal ultrasound and reported an 11.3 mm/year growth rate detected 13 of 16 malignancies (carcinoma in situ and invasive carcinoma) with an 81 % accuracy [10]. A more recent study by Kang [6] examined 201 patients with low-risk BD-IPMN smaller than 30 mm with median follow-up of 28 months. The study reported a higher 5-year risk of malignancy for BD-IPMNs growing faster than 2 mm/year (44.1 vs. 0.8 %). The aim of this study was to assess whether growth rates of suspected BD-IPMN could identify BD-IPMNs at higher risk of developing malignancy. Our study population consisted of 284 patients with presumptive BD-IPMN without high risk or worrisome features that would be recommended for imaging surveillance consistent with the international consensus guidelines [1] and AGA guidelines [2]. We found a significant association between BD-IPMN growth rate and development of invasive carcinoma. A BD-IPMN growth rate between 2 and 5 mm/year is associated with an increased risk of malignancy with a hazard ratio of 11.4 compared to BD-IPMN growth rates less than 2 mm/year. BD-IPMNs with a growth rate C5 mm/year
Dig Dis Sci (2015) 60:2800–2806
have an increased risk of malignancy with a higher hazard ratio of 19.5 compared to BD-IPMN growth rate less than 2 mm/year. Based on our results, BD-IPMN total growth and growth rates appear promising in their ability to identify BD-IPMN at high risk of malignancy. However, there are limitations that should be noted. From a pragmatic standpoint, the ability to measure a 5-mm difference on imaging is likely to be more reliable than measuring a 2-mm difference. A prior study demonstrated that variability in pancreatic cyst size measurements between CT, MRI, and EUS studies is 3 to 4 mm [11]. Attempts to detect a 2-mm difference on imaging are more vulnerable to human errors in measurement, measurement of the cyst at different levels, and differences in imaging modalities. A 5 mm/year growth would be more reliably detected on imaging, has a higher hazard ratio for malignancy, and higher specificity for malignant transformation. A higher specificity is desirable when evaluating BD-IPMN for malignancy to avoid unnecessary resections. A 2 mm/year growth rate over two or more years would produce at least a 4-mm difference, which could be more reliably detected. Both 2 and 5 mm/ year cutoffs demonstrate a high negative predictive value of 99 % that could be utilized to reassure patients with slow growing BD-IPMNs regarding the benign nature of their cyst and also reduce the number of unnecessary resections. Another limitation of this study is the retrospective nature. Surveillance was not standardized as this was a retrospective study across four centers. Follow-up imaging was generally recommended on at least a yearly basis, though patients and referring physicians did not necessarily adhere to recommendations. The study also based the diagnosis of BD-IPMN on clinical and imaging characteristics, rather than surgical resection. However, surgical resection would have eliminated our ability to follow the natural history and growth of these cysts. The diagnosis of BD-IPMN based on clinical and EUS evaluation represents actual clinical practice. Routine cyst fluid aspiration for CEA and serum measurement of CA 19–9 was not performed in all cases as guidelines do not recommend these as part of routine evaluation. There was also potential selection bias as our study population was referred for EUS evaluation. However, despite a bias toward a higher-risk population, the overall rate of BD-IPMN malignancy was 3 %, which is consistent with prior studies [6] that also excluded worrisome and high-risk BD-IPMN. The patients in the study were followed clinically for a median 56 months and with imaging surveillance for 36 months. The imaging surveillance was shorter due to several reasons including the discontinuation of imaging surveillance due to stability of the cyst or the patient was no longer deemed a surgical candidate (due to age or
2805
comorbidities), or personal preference to discontinue surveillance. Overall, the length of both clinical and imaging surveillance is a strength of the study, allowing for the capture of the long-term natural history and malignant transformation of BD-IPMN. Resection is likely not necessary in all cases of BDIPMN enlarging faster than 2 mm/year, but these patients should be evaluated with EUS and receive closer surveillance. The 2 and 5 mm yearly growth rates can be followed over the course of multiple years (as opposed to just 1 year) as was done in this study which may produce a more reliable trend and indicator of malignancy. Should the cyst continue to exhibit significant growth or develop other worrisome features, then resection should be considered. We also found that a high percentage of patients with BD-IPMN that doubled in size or grew more than 10 mm since initial detection have a high rate of malignancy (50 and 36 %, respectively). A 40 % increase in cyst size has a sensitivity, specificity, and accuracy of 78, 84, and 84 %, respectively, for the detection of malignancy. The 2 mm/year growth rate and 40 % percentage growth both perform favorably in their ability to identify malignant BD-IPMN (Table 3) compared to the current consensus guidelines, which demonstrate a reported sensitivity, specificity, and accuracy of 72, 78, and 80 %, respectively [3]. In conclusion, in this multicenter study of patients with suspected BD-IPMN undergoing long-term surveillance imaging, we demonstrate that BD-IPMN growth rates greater than 2 and 5 mm/year are associated with an increased risk of malignant transformation. In addition, total BD-IPMN growth was also associated with an increased risk of malignancy with all malignant BD-IPMNs growing at least 10 mm prior to a cancer diagnosis. Suspected BDIPMN growth rates of 2 or 5 mm/year or a total growth of 10 mm should be considered for inclusion in future guidelines as worrisome features for malignant transformation of BD-IPMN. Perhaps of equal or greater value is the ability to reassure patients that BD-IPMN growing less than 2 mm/year have less than 1 % rate of malignancy during long-term surveillance and minimize unnecessary resections. Acknowledgments The study was partially supported by the National Institutes of Health Grant UL1TR000100 (Ronghui Xu). Conflict of interest
None.
References 1. Tanaka M, Ferna´ndez-del Castillo C, Adsay V, et al. International consensus guidelines 2012 for the management of IPMN and MCN of the pancreas. Pancreatology. 2012;12:183–197.
123
2806 2. Vege S, Ziring B, Jain R, et al. American gastroenterological association institute guideline on the diagnosis and management of asymptomatic neoplastic pancreatic cysts. Gastroenterology. 2015;148:819–822. 3. Jang JY, Park T, Lee S. Validation of international consensus guidelines for the resection of branch duct-type intraductal papillary mucinous neoplasms. Br J Surg. 2014;101:686–692. 4. Kim KW, Park SH, Pyo J, et al. Imaging features to distinguish malignant and benign branch-duct type intraductal papillary mucinous neoplasms of the pancreas; a meta-analysis. Ann Surg. 2014;259:72–81. 5. Shimizu Y, Yamaue H, Maguchi H, et al. Predictors of malignancy in intraductal papillary mucinous neoplasm of the pancreas: analysis of 310 pancreatic resection patients at multiple high-volume centers. Pancreas. 2013;42:883–888. 6. Kang MJ, Jang JY, Kim SJ, et al. Cyst growth rate predicts malignancy in patients with branch duct intraductal papillary mucinous neoplasms. Clin Gastroenterol Hepatol. 2011;9:87–93.
123
Dig Dis Sci (2015) 60:2800–2806 7. Bae SY, Lee KT, Lee JH, Lee JK, Lee KH, Rhee JC. Proper management and follow-up strategy of branch duct intraductal papillary mucinous neoplasms of the pancreas. Dig Liver Dis. 2012;44:257–260. 8. R Core Team. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2014. 9. Therneau T (2014) A package for survival analysis in S. R package version 2.37-7. 10. Yamaguchi T, Baba T, Ishihara T, et al. Long-term follow-up of intraductal papillary mucinous neoplasm of the pancreas with ultrasonography. Clin Gastroenterol Hepatol. 2005;3:1136–1143. 11. Maimone S, Agrawal D, Pollack MJ, et al. Variability in measurements of pancreatic cyst size among EUS, CT, and magnetic resonance imaging modalities. Gastrointest Endosc. 2010;71: 945–950.
