Schlüsselwörter Pankreatische Zysten - Neoplasie - zystisch - muzinös - serös - Bildgebung - Endosonografie
- EUS
Keywords Pancreatic cysts - Neoplasms - Cystic - Mucinous - Serous - Diagnostic Imaging - Endosonography
- EUS
Introduction
Cystic pancreatic lesions (CPL) are frequently found in transabdominal ultrasound,
CT, or MRI scans of the abdomen [1 ]. Prevalence data on CPL range from 2.4% [2 ] to 49.1% in MRI, increase with age [3 ] and better resolution capacity due to recent developments in imaging techniques
[4 ]. Our study focused on the five most frequent lesions, comprising 95% of all CPL
that include: pseudocysts, intraductal papillary mucinous neoplasms (IPMN), mucinous
cystic neoplasms (MCN), serous cystic neoplasms (SCN), and solid pseudopapillary neoplasms
(SPN). Due to their premalignant potential the latter four are designated as cystic
pancreatic neoplasms (CPN).
Pancreatitis-associated pseudocysts are the most often detected CPL (30%) [5 ] and develop when peripancreatic lipid necroses are resorbed. There is no danger
of malignant progression [6 ].
IPMN as mucine-producing neoplasia account for about 20% of CPL [5 ], and can be divided into main duct (MD-IPMN), branch duct (BD-IPMN) and mixed type
IPMN implicating different prognoses [7 ]. In 2017, a meta-analysis on BD-IPMN reported a probability of 0.98% per patient
year for malignant transformation [8 ]. In contrast, in resected MD-IPMN, 38% to 68% of cases exhibited high grade dysplasia
or even carcinoma [9 ].
Cystadenomata can be differentiated into SCN and MCN. MCN are found in 10% of CPL,
show a female preponderance, do not communicate with the pancreatic duct, and in most
cases are found in the pancreatic tail. High-grade dysplasia or carcinoma are seen
in 9.9% of patients, often showing sizes > 60 mm and mural nodules [10 ]. About 20% of CPL are SCN, 75% affecting the female sex, and are characterized by
a cluster of small cysts often resembling honeycomb. SCNs are most often asymptomatic
and grow slowly with a minimal risk of malignant transformation [11 ].
SPN (5% of CPL), 90% affecting women, show a mean size of 8.6 cm (SD +/–4.3), and
can often be well demarcated from normal pancreatic tissue by pseudopapillary and
pseudo-cystic parts [12 ]. Despite their malignant potential with metastases in 7.7% to 19% of patients, SPN
generally show a good prognosis [13 ].
Due to the low specificity of morphological signs, endoscopic ultrasound (EUS) assignment
of CPNs to one of the described entities can be challenging, and often is only feasible
when clinical data and/or other imaging techniques are integrated. Of central importance
is the differentiation in serous and mucinous cysts as the latter exhibit a risk for
malignant transformation [14 ]. Analysis of carcinoembryonic antigen (CEA) in cyst content can help classify cyst
entity; however, CEA is not capable of differentiating IPMN from MCN or high-grade
dysplasia from carcinoma [15 ].
Bearing in mind the increasing rate of random diagnoses of CPL, the questions arise
of how to adequately supervise particular patients and how to balance the risk of
malignant development versus the implication of surgery with consecutive post-surgical
restrictions patients often have to face.
Material and Methods
For our retrospective study, data on EUS examinations with first diagnosis of a CPL
were analyzed from January 1st 2012 to December 31st 2018. Patients and EUS data were extracted from our electronical reporting system
(ClinicWinData, version 8.08.03.) in the Hospital Barmherzige Brüder Regensburg, Germany.
EUS had been performed by 4 experienced endoscopic physicians with at least 5 years
of experience using radial and linear endoscopes [Pentax], and a EUS processor Preirus
(Hitachi). Diagnostic criteria are defined as outlined in the European guidelines
on pancreatic cystic neoplasms [16 ]. Diagnoses of CPN were made by referring to the last EUS diagnosis in the patients’
charts; thus, also information on differences in size over time, changes in cystic
structure, results of fine needle aspiration (FNA) or complementary imaging could
have been taken into account. FNA was used to further delineate the character and
dignity of cystic processes, and was performed using a 22G needle. Extracted material
was analyzed biochemically (CEA, lipase), cytologically (serous or mucinous fluid)
and/or histologically. Histopathological results after surgery enabled a comparison
of EUS diagnoses and served as gold standard for correctness of EUS.
Institutional Review Board approval for the study was obtained from the University
of Regensburg (ethical committee No. 20-1948-104).
Statistical analysis was performed using “IBM SPSS statistics version 25”. For analysis
of change in size, first and last examination were used. Statistical analysis comprised
Chi-square test in larger groups, Fisher’s-exact test in smaller groups, and Mann-Whitney
U-test. A difference was considered significant at p < 0.05.
Results
EUS diagnoses in CPN
After exclusion of all directly malignoma-suspect findings, 455 patients with 798
examinations were diagnosed with a CPL. In 160 patients (35.2%) the CPLs could not
be categorized as CPN (no wall irregularities, no relation to pancreatic duct, “harmless
cyst”) and were excluded from further analysis. 72 patients exhibited pseudocysts.
The remaining group of CPN included 223 patients involving 390 examinations ([Fig. 1 ]). BD-IPMN was diagnosed in 138 patients, and was the most frequent diagnosis in
our study collective. MD-IPMN diagnosis was made in 16 patients, mixed-type IPMN in
5 patients. 60 patients were diagnosed with cystadenoma, classified as SCN in 46 patients
and MCN in 6. In 8 patients, the cystadenoma could not be sub-specified (= unclassifiable).
SPN were seen in 4 patients. Epidemiological data are summarized in [Table 1 ].
Fig. 1 Stratification of patients with cystic pancreatic lesions using endoscopic ultrasound.
Table 1 Patientsʼ characteristics in the CPN study population.
Patients
Number
Sex [f/m]
Median age (range) [years]
Abbreviations: CPN=cystic pancreatic neoplasm; IPMN=intraductal papillary mucinous
neoplasms; MD-IPMN=main duct IPMN; BD-IPMN=branch duct IPMN; Mt-IPMN= mixed type IPMN;
SCN=serous cystic neoplasms; MCN=mucinous cystic neoplasms; SPN=solid pseudopapillary
neoplasms.
All CPN patients
223
146/77
69 (15–95)
BD-IPMN
138
91/47
71 (34–95)
MD-IPMN
16
11/5
76 (54–88)
Mt-IPMN
5
2/3
79 (73–88)
SCN
46
19/27
67 (21–81)
MCN
6
5/1
67 (46–74)
Unclassifiable cystadenoma
8
8/–
76 (62–82)
SPN
4
2/2
30 (15–42)
Size progression
Follow-up time was defined as the period of time between first and last documented
EUS. 98 patients with more than one examination had a median follow-up time of 11.7
months (range 1 to 41), and a mean of 2.7±1.05 examinations per patient ([Fig. 2 ]). Progression in cystic size was seen in 20/52 patients with BD-IPMN, but in only
6/20 patients size increase measured >2 mm. 2/5 patients with MCN and 6/21 SCN patients
showed increases in size, while in the latter only 4/6 size increases >2 mm were detected.
In all other patients, no progression in cystic size could be observed during the
follow-up period.
Fig. 2 Time of follow-up in patients with different CPNs.
Fine needle aspiration and cytological analysis
FNA was performed in 21 patients. In 14/21 patients, FNA supported the diagnosis of
SCN (serous fluid or inconsistent findings arguing against a mucinous process, CEA
<192 ng/ml). In 1 out of 3 suspected MCN, 2 MD-IPMN and 1 BD-IPMN patients, mucinous
fluid was detected. Overall, cytological analysis was obtained in 19 patients but
was able to differentiate between serous and mucinous only in 9 patients. In 13/16
patients, CEA levels were below the cut-off value of 192 ng/mL, underscoring the rather
benign character of the lesion. Taken together, FNA ± cytological/CEA analysis confirmed
the suspected diagnosis in 12/21 patients, and enabled an alternative diagnosis in
9/21 patients. For further detail see [Table 2 ].
Table 2 Characterization of cystic pancreatic neoplasms in comparison before and after FNA
using lipase concentration, CEA concentration and cytological analysis. Numbers in
bold represent the decisive parameter(s) for diagnosis after having performed FNA.
Number of patients
Diagnosis before FNA
Lipase
CEA
Cytology
Diagnosis after FNA
[µ/l]
<cut-off
>cut-off
not available
serous
mucinous
not indicative
not available
Abbreviations: IPMN=intraductal papillary mucinous neoplasms; MD-IPMN=main duct IPMN;
BD-IPMN=branch duct IPMN; MCN=mucinous cystic neoplasms; SCN=serous cystic neoplasms;
SPN=solid pseudopapillary neoplasms; CEA= carcinoembryonic antigen; FNA=fine needle
aspiration.
8
SCN
8.8; SD 5,5
(n = 2)
5
0
3
2
0
5
1
8 SCN
1
BD-IPMN
695.3
(n = 1)
1
0
0
0
0
1
0
1 SCN
4
Unspecified
60726.8;
SD 60633,2
(n = 2)
2
1
1
2
0
2
0
4 SCN
1
Cystadenoma
20.9
(n = 1)
0
1
0
0
0
1
0
1 SCN
1
Unspecified
36865.0
(n = 1)
1
0
0
0
0
1
0
1 MCN
2
MCN
–
2
0
0
0
1
0
1
2 MCN
2
MD-IPMN
–
1
0
1
0
2
0
0
2 MD-IPMN
2
SCN
736259.7
(n = 1)
1
1
0
1
1
0
0
2 BD-IPMN
21
–
13
3
5
5
4
10
2
Indication for surgery
37 patients were listed for surgery. According to the European Consensus Guidelines,
8/37 patients fulfilled absolute criteria for surgery: jaundice in 5 patients, main
duct (MD) dilatation in 4 patients. Other absolute indications for surgery like positive
cytology for malignancy, high grade dysplasia, a solid mass, or mural nodules ≥ 5
mm were not detected. 22/37 patients (9.4% of all 223 patients with CPN) listed for
surgery were diagnosed with ≥ 1 relative criteria for surgery (one patient exhibited
criteria of both absolute and relative indications): 8 with cyst diameter ≥ 40 mm,
6 with MD dilatation 5–9.9 mm, 7 with acute, and 13 with chronic pancreatitis.
8 of these 37 patients were listed for surgery without fulfilling the European Consensus
Guidelines’ criteria for surgery. For further detail see [Table 3 ].
Table 3 Reasons for resection in 8 patients without fulfilling European evidence-based guidelines
on pancreatic cystic neoplasms, post-surgery diagnoses.
Patient
Age [years], sex
Reason for resection
Resection performed?
Post-surgery diagnosis
Abbreviations: EUS=endoscopic ultrasound; IPMN=intraductal papillary mucinous neoplasms;
MD-IPMN=main duct IPMN; MCN=mucinous cystic neoplasms; SCN=serous cystic neoplasms.
1
43, female
size progressive cyst within one year – no pre-EUS documented
yes
SCN
2
54, male
MD-IPMN
yes
SCN
3
68, male
MCN and size progression of 4 mm in two years
no resection, only indication
–
4
58, male
no differentiation between MCN and IPMN possible
yes
ductal adenocarcinoma
5
21, female
polycystic SCN of 10 cm size without single cyst >40 mm
yes
pancreatic pseudocyst
6
64, male
non-classifiable CPN after having used different imaging techniques
no resection due to intra-surgery
ultrasound finding
–
7
84, male
polycystic CPN of 6 cm size without single cyst >40 mm
no resection, only indication
–
8
25, male
suspicion of non-classifiable cystadenoma, indication because of young age
no resection, only indication
–
Whereas surgical resection was indicated already in 23/28 patients after the first
EUS examination, surgery was indicated in only 5 patients during follow-up (data not
shown).
Comparison EUS and post-surgery diagnoses
(Partial) pancreatectomy was performed in 28 patients. In 10 of 28 patients (35.7%)
EUS diagnosis was correct: 3 pseudocysts, one SPN, 6 MD-IPMN. The highest rate of
correct classification was found in the group of MD-IPMN patients (6/7 patients, 85.7%),
5 of whom had exhibited absolute risk criteria, one patient a relative risk criterion.
The remaining assumed MD-IPMN lesion was diagnosed as SCN without risk of malignant
transformation. EUS differentiation into serous and mucinous CPNs when compared to
histopathological diagnosis after surgery was correct in 13/19 patients (68.4%; [Table 4 ]). In 8/28 patients, EUS had not been able to categorize the CPN, 4 of which constituted
malignant tumors. In the remaining 10 patients EUS diagnosis turned out to be not
correct (for details see [Table 5 ]). 9 patients did not undergo surgery due to individual reasons ([Table 6 ]).
Table 4 EUS diagnosis of serous and mucinous cystic pancreatic neoplasms when compared to
histopathological diagnosis after surgery.
EUS diagnosis
n
Correct
Not correct
Diagnosis after surgery
Abbreviations: EUS=endoscopic ultrasound; IPMN=intraductal papillary mucinous neoplasms;
MD-IPMN=main duct IPMN; MCN=mucinous cystic neoplasms; SCN=serous cystic neoplasms;
SPN=solid pseudopapillary neoplasms.
pseudocyst
4
3
1
intrapancreatic mucinous neoplasia
MD-IPMN
7
6
1
SCN
MCN
2
2
1 “no findings”, 1 pseudocyst
SCN
2
1
1
retroperitoneal neurinoma
SPN
4
3
1
MCN
Total
19
13
6
Table 5 Incorrect EUS diagnoses in 10 patients after comparison with post-surgery diagnosis.
EUS diagnosis
Post-surgery diagnosis
Abbreviations: EUS=endoscopic ultrasound; IPMN=intraductal papillary mucinous neoplasms;
MD-IPMN=main duct IPMN; MCN=mucinous cystic neoplasms; SCN=serous cystic neoplasms;
SPN=solid pseudopapillary neoplasms.
1
pseudocyst
intrapancreatic mucinous neoplasia
2
MD-IPMN
SCN
3
unclassifiable cystadenoma
pseudocyst
4
MCN
no pathological finding
5
MCN
MD-IPMN
6
SCN
pseudocyst
7
SCN
retroperitoneal neurinoma
8
SPN
chronic calcifying pancreatitis
9
SPN
chronic calcifying pancreatitis
10
SPN
MCN
Table 6 9 patients with indication for surgery but actually not undergone, reasons as found
in patients’ charts.
Age
Description of CPN
No surgery – reasons
Abbreviations: CPN=cystic pancreatic neoplasia; IPMN=intraductal papillary mucinous
neoplasms; BD-IPMN=branch duct IPMN; MCN=mucinous cystic neoplasms; SCN=serous cystic
neoplasms; SPN=solid pseudopapillary neoplasms.
47
CPN non-classifiable
during follow-up diagnosis of pseudocyst
78
BD-IPMN
icterus presumably caused by cholelithiasis
68
MCN
using FNA diagnosis of only low-grade dysplasia
69
SCN
during follow-up diagnosis of pancreatic carcinoma (different localization)
53
CPN non-classifiable
explorative laparotomy with biopsy, no malignoma
78
CPN non-classifiable
surgery denied by patient
64
CPN non-classifiable
due to intraoperative ultrasound no curative surgery possible any more
84
CPN non-classifiable
due to age and multimorbidity
25
Cystadenoma non-classifiable
lost to follow-up
Overall, 7 of 28 resected CPNs (25%) showed malignant transformation (3 patients with
MD-IPMN, 4 patients with unspecified findings in EUS: 2 ductal adenocarcinomata, 1
neuroendocrine carcinoma, and 1 neuroendocrine tumor with suspected malignancy). Fisher’s-Exact-Test
showed no correlation between positive absolute criteria for surgery and malignancy
(χ²(1. N = 28) = 2.455, p = 0.117). Relative criteria for surgery did not correlate with malignancy either
(χ²(1, N = 28) = 0,474 p = 0,491).
Discussion
Our study was a retrospective analysis of CPLs (798 patients) with focus on CPNs (223
patients) in a tertiary referral center in Germany. The high percentage of IPMN and
the low portion of MCN is of note: IPMN were diagnosed in 159 patients of which 16
were MD-IPMN (7.2%), five mixed-type IPMN (2.2%) and 138 BD-IPMN (61.9%). These frequencies
are similar to a recent survey reporting numbers of 4.6, 6.2, and 70.1%, respectively
[16 ], but contrast with a study by Sahora et al. [17 ]. Differing frequencies of diagnoses between observational studies and studies investigating
resected CPNs could be caused by selection bias on the one hand; since in 90/138 patients
with BD-IPMN (65.2%) no clear communication with the main pancreatic duct could be
demonstrated, on the other hand this fact could open the possibility that some of
these patients potentially might have had another pathology like MCN. After all, according
to recent guidelines, this differentiation appears to be difficult [18 ]. Following this possibility, of 60 cystadenomata only 6 were defined as MCN. This
small number of patients with differing epidemiological data when compared to the
literature (median age 68 years; thus, far higher than 45 to 48 years; lower female
percentage of 83.3% when compared to 95% [19 ] to 99.7% [11 ]
[14 ]; median cystic size with 3.5 cm smaller than 5 to 8.7 cm [14 ]
[20 ]; localization less often in pancreas corpus or tail (50% versus up to 97% [14 ]
[21 ], respectively) might underline the explanation given above.
Whereas patients with suspected MD-IPMN, MCN, and SPN diagnoses most often underwent
surgery, BD-IPMN and SCN patients were most likely to have EUS follow-up. In 38.5%
of patients with BD-IPMN cystic size increased over time which is comparable to the
literature ranging from 29% [21 ] to 54.2% [22 ]. Of note is that depending on the study, the definition of when a cyst is classified
as size progressive is different, as is the method of measurement [23 ]
[24 ]. In our study, any change of size was defined as either size increasing or decreasing
and might therefore explain differences compared to other studies. When applying a
cut-off of 2 mm, in only 11.5 % of BD-IPMN patients was size progression detected.
In patients with SCN, when applying the cut-off value of 2 mm of size change [25 ], only seven patients would have qualified for a relevant size progression. Due to
these low frequencies of relevant size progression, no BD-IPMN and only two SCN patients
underwent surgery.
Size progression is variably linked to an enhanced risk of malignancy; whereas some
studies could not detect a difference in the rate of malignancy between size progressors
and non-progressors [24 ]
[25 ], Akahoshi et al. found a higher risk of malignancy for a cut-off value of 3.5 mm/year
with a sensitivity of 88%, a specificity of 91%, and a precision of 93% [26 ]. Whereas European guidelines define a size progression of 5 mm/year as relative
surgery indication [19 ], this criterion could not be found in our study collective – cut-offs of 2 mm or
3.5 mm would have applied for only three patients each, respectively. Within the time
of supervision over 3.5 years, no malignant transformation in our BD-IPMN patients
was documented. Bearing in mind the risk of malignant transformation of 0.65–0.8%
per year as described in a systematic review in 2017 and a cumulative incidence of
pancreatic carcinoma of 7.77% in ten years, our results are well in line with the
literature.
Only 21 patients (17 presumed cystadenomata, among which 14 SCN; two BD-IPMN and two
MD-IPMN patients) underwent FNA due to suggestive (pre)malignant signs or size progression.
Since there were no evident differences between the 4 investigators with respect to
the indication for FNA, this relatively small proportion also reflects the low proportion
of size progressors and/or missing suspect signs of the cystic lesion. In 12 cases,
cytological results underscored the presumed diagnosis, in 16 cases CEA levels could
be determined, the latter reflecting with 76.2% a slightly higher proportion of patients
with available CEA when compared to studies by Winner (72,5%; [21 ]) and Khalid (67%; [25 ]). For classification into mucinous and serous cysts the cut-off of 192 ng/mL was
used as recommended by European guidelines. In 8 patients, CEA levels and cytological
analysis were not coherent; this contrasting finding is not unknown and has led to
an algorithm published in 2004 classifying cysts as mucinous if one of the parameters
morphology, cytology, or CEA levels are positive resulting in a higher sensitivity
of 91% but lower specificity [26 ]. Unfortunately, adoption of this proposed algorithm would not have led to more coherent
results either.
It is common sense that CEA levels are not capable of distinguishing pre-malignant
and malignant mucinous lesions [27 ]
[28 ]. In line with this difficulty, 4 patients underwent surgical resection after FNA:
in neither the patients with very low nor the patient with extremely high CEA levels
did the lesions histologically turn out to be malignant. The largest prospective study
with 198 patients so far investigating the quality of FNA proposed the combination
of CEA and cytological analysis to better estimate malignancy, and actually identified
lesions correctly in only 45.8% of cases as malignant [29 ]. Compared with this study, recall ratio between mucinous and non-mucinous lesions
in our study was 68.4%.
In 7 MD-IPMN patients, surgery was executed, in 85.7% the EUS diagnosis turned out
to be correct. The percentage of malignancy was 42.8%, and, thus, smaller than in
a study by Hinz et al., who found malignancy in 55.8% of patients [29 ]. In this study, malignancy of 33% was true in patients exhibiting a pancreatic duct
of 5–9.9 mm, whereas malignancy could be demonstrated in 67% with a main duct of ≥
9.9 mm [30 ]. Due to the small number of patients with this diagnosis in our study, stratification
according to the main duct diameter could not be performed. In only 1/4 patients with
suspected SPN this diagnosis was actually histologically proven. Even though the presumed
diagnosis of SPN was rather discussed as differential diagnosis in these 4 cases the
fact that only one turned out to be correct implies a difficult EUS diagnosis. In
line with this presumption, Yu et al. reported in a systematic review in 2010 on only
23.7% correct SPN diagnoses in 325 histologically proven SPN lesions [31 ]. In contrast, Tjaden et al. showed in 70% a correct SPN pre-surgery diagnosis in
2019 [32 ]. In turn, Liu et al. reported in 2019 on relatively unspecific EUS findings in SPN
lesions [30 ].
The proportion of patients in our study in which EUS diagnosis could be correlated
with the gold standard histopathology was in fact low (surgical resection indicated
in 37 patients (8.1%) and – due to individual reasons – actually performed in only
28 patients). Of note, in only 7 patients, malignoma was diagnosed.
With respect to 20 defined diagnoses presumed by EUS, 10 were correct (50%). If 8
further patients in which our investigators had not been able to specify a diagnosis
via EUS pre-surgery were included in the analysis, 35.7% of the diagnoses were correct.
In comparison, a study performed to differentiate SCN, MCN, and cystadenocarcinoma
reported on correct assignment in 34%, 28%, and 22%, respectively [33 ].When trying to classify into mucinous or non-mucinous, accuracy to diagnose mucinous
CPL was 68.4%, and, thus, in the same range as in a prospective study by de Jong et
al. (75%; [32 ]). Again, better values in our study could potentially be explained by additional
information accessible to our investigators.
Limitations of the study
Our patient collective was not representative of the general population. A loss to
follow-up bias could not be excluded. Furthermore, investigators in our study could
use results from other imaging studies leading potentially to a higher hit rate of
EUS with respect to the correct diagnosis. Since results were analyzed according to
prevailing European guidelines, parameters such as CA 19-9 in serum or newly diagnosed
diabetes mellitus were not part of the analysis. Moreover, contrast enhanced ultrasound
(CEUS) was not performed on a regular basis. Analysis of size variability was restricted
to the largest cyst, and, therefore, other pancreatic cysts were neglected. In total,
only few patients underwent FNA (21/223).
Exclusion of 160 patients with non-defined entities, assignment of cysts to seven
entities, and a low frequency of these distinct CPNs rendered statistical analysis
difficult, despite the initial relatively high number of 455 patients. In contrast,
inclusion of any defined CPNs to the analysis enabled a robust overview on the frequency
of CPNs in a third level reference center in Germany. While studies so far have focused
solely on distinct patient groups such as specific CPNs, defined risk criteria, FNA,
or resections, our study involved all of the mentioned parameters; furthermore, our
analysis also comprised the course of distinct CPNs over time. Our study, thus, enabled
a comprehensive overview of frequency, diagnostic correctness, and course over time
of distinct CPNs in clinical routine.
Conclusions
Size progression in CPN during follow-up is rather rare, and most frequently, not
clinically relevant. Detection of malignancy is seldom. Even in a tertiary reference
center and following European guidelines, EUS accuracy in delineating diagnosis and
dignity of CPNs is rather low. The data of this study suggest that morphologic criteria
to assess pancreatic cysts alone are not sufficient to allow a clear diagnosis. Hence,
for the improved assessment of pancreatic cysts, EUS should be combined with additional
tests and techniques such as MRT/MRCP, contrast-enhanced EUS, and/or FNA/fine needle
biopsy including fluid analysis. The combination and correlation of imaging studies
with EUS findings is mandatory.
