ORIGINAL ARTICLE
Hepatobiliary Mucinous Cystic Neoplasms With Ovarian Type Stroma (So-Called “Hepatobiliary Cystadenoma/Cystadenocarcinoma”) Clinicopathologic Analysis of 36 Cases Illustrates Rarity of Carcinomatous Change Brian Quigley, MD,* Michelle D. Reid, MD,* Burcin Pehlivanoglu, MD,* Malcolm H. Squires, III, MD,† Shishir Maithel, MD,† Yue Xue, MD, PhD,* Choi Hyejeong, MD,‡ Gizem Akkas, MD,* Takashi Muraki, MD,* David A. Kooby, MD,† Juan M. Sarmiento, MD,† Ken Cardona, MD,† Aarti Sekhar, MD,§ Alyssa Krasinskas, MD,* and Volkan Adsay, MD8
Abstract: The literature is highly conflicting on hepatobiliary mucinous cystic neoplasms (MCNs), aka “hepatobiliary cystadenoma/ cystadenocarcinoma,” largely because ovarian stroma (OS) was not a requirement until WHO-2010 and is not widely applied even today. In this study, MCNs (with OS) accounted for 24 of 229 (11%) resected hepatic cysts in one institution. Eight of the 32 (25%) cysts that had been originally designated as hepatobiliary cystadenoma/ cystadenocarcinoma at the time of diagnosis proved not to have an OS during this review and were thus re-classified as non-MCN. In total, 36 MCNs (with OS) were analyzed—24 from the institutional files and 12 consultation cases. All were women. Mean age was 51 (28 to 76 y). Mean size was 11 cm (5 to 23 cm). Most (91%) were intrahepatic and in the left lobe (72%). Preoperative imaging mentioned “neoplasm” in 14 (47%) and carcinoma was a differential in 6 (19%) but only 2 proved to have carcinoma. Microscopically, only 47% demonstrated diffuse OS ( > 75% of the cyst wall/lining); OS was often focal. The cyst lining was often composed of non-mucinous biliary epithelium, and this was predominant in 50% of the cases. Degenerative changes of variable amount were seen in most cases. In situ and invasive carcinoma was seen in only 2 cases (6%), both with small invasion (7 and 8 mm). Five cases had persistence/ recurrence, 2 confirmed operatively (at 7 mo and 15 y). Of the 2 cases with carcinoma, one had “residual cyst or hematoma” by radiology at 4 months, and the other was without disease at 3 years. In conclusion, many cysts (25%) previously reported as hepatobiliary cystadenoma/cystadenocarcinoma are not MCNs. True MCNs are uncommon among resected hepatic cysts (11%), occur exclusively in females, are large, mostly intrahepatic and in the From the Departments of *Pathology; †Surgery; §Radiology, School of Medicine, Emory University, Atlanta, GA; ‡Department of Pathology, Ulsan University College of Medicine, Ulsan, Korea; and 8Department of Pathology, Medical College of Wisconsin Milwaukee, WI, USA. Conflicts of Interest and Source of Funding: The authors have disclosed that they have no significant relationships with, or financial interest in, any commercial companies pertaining to this article. Correspondence: Volkan Adsay, MD, Department of Pathology, Medical College of Wisconsin Milwaukee, WI 53226 (e-mail: [email protected]). Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Am J Surg Pathol
Volume 00, Number 00, ’’ 2017
left lobe (72%). Invasive carcinomas are small and uncommon (6%) compared with their pancreatic counterpart (16%). Recurrences are not uncommon following incomplete excision. Key Words: mucinous cystic neoplasm, hepatobiliary cystadenoma, hepatobiliary cystadenocarcinoma, liver cyst, hepatic cyst (Am J Surg Pathol 2017;00:000–000)
hen “hepatobiliary cystadenoma/cystadenocarcinoma” was first presented to the literature as an entity, the category was interpreted to embrace essentially any epitheliallined cystic lesion occurring in the liver and biliary tract.1–4 As more knowledge accumulated about the nature and classification of cystic lesions in the hepato-pancreatobiliary tract, the view on the criteria needed to make this diagnosis in the liver is now changing dramatically, many of them adapted from the pancreas. In the 1980s, essentially 2 types of cystic tumors had been characterized in the pancreas: mucinous and serous. Over the years, while the definitions of serous tumors remained unchanged,5 those for the “mucinous” category have undergone major modifications. First, cyst-forming intraductal tumors (lined by ductal/mucinous cells), namely, intraductal papillary mucinous neoplasms, intraductal oncocytic papillary neoplasms, and intraductal tubulopapillary neoplasms, were characterized.6–16 More recently simple cysts17 and other mimickers such as secondary cystic dilatations, congenital processes, and others have been delineated. The most important aspect of this paradigm shift in the pancreas that was pertinent to hepatic cysts was that, it has become clear that the presence of ovarian stroma delineates a pathologically and biologically distinct cyst type, and therefore, WHO established ovarian stroma (OS) as a requirement for the diagnosis of mucinous cystic neoplasm (MCN), both for the pancreas and liver. This stroma has all the histologic and immunophenotypic
W
www.ajsp.com
|1
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Quigley et al
Am J Surg Pathol
characteristics of ovarian cortical stroma, including the presence of luteal type cells. Defined by OS, in the pancreas MCN has very distinctive features including almost exclusive ( > 98%) occurrence in women and vast majority in the body/tail as a well-demarcated cystic mass,18 which is substantially different than the earlier impressions on this entity19–22 before the OS requirement. In fact, there are still publications without histologic verification that lead to a different clinical picture.23 Once defined by OS, resected pancreatic MCNs harbor invasive carcinoma in 16%, mostly of tubular type but occasionally of the sarcomatoid/undifferentiated type.18,24–27 The criteria established for the classification of cystic tumors in the pancreas is now also adopted to the hepatobiliary tract in which the nonspecific and tainted diagnosis of “hepatobiliary cystadenoma/cystadenocarcinoma” as an entity has been eliminated by the WHO-2010, and replaced by the more specific entities including MCN (defined by the presence of OS) as well as hepatobiliary counterparts of cystic intraductal tumors, namely intraductal papillary neoplasms, intraductal tubulopapillary neoplasms, and intraductal oncocytic papillary neoplasms. It is also being appreciated that not all biliary-lined cysts are MCNs, and that there is radiographic overlap with cystic bile duct hamartomas, congenital cysts, and Caroli disease. However, most of the current impressions in the literature about cystic hepatobiliary lesions are still based on the earlier criteria and literature of hepatobiliary cystadenoma/cystadenocarcinoma.23 The purpose of this study was to elucidate the relative frequency of hepatobiliary MCNs (as now defined by the WHO by the presence of OS) among hepatic cysts; determine its clinicopathologic characteristics and establish the true frequency of carcinomatous changes in this tumor type.
MATERIALS AND METHODS Appropriate institutional approvals were obtained for this retrospective review.
Case Selection All 24 tumors that had been designated as hepatic or biliary cystadenoma or cystadenocarcinoma or mucinous cystic neoplasm in the authors’ institutional files were retrieved and examined. Separately, 229 hepatic cystic lesions resected specifically in the authors’ institution between the years of 1994 and 2014 were reviewed specifically for the presence or absence of ovarian stroma. Only cases with unequivocal ovarian stroma —comprising densely packed spindle cells with elongated nuclei and scant cytoplasm underneath the epithelium—were included in the study. In cases where the ovarian stroma was focal or less cellular (eg, atrophic), its nature was confirmed with immunohistochemical staining for progesterone receptor (PR), estrogen receptor (ER) and inhibin. In addition, pathology consultation files of the authors were screened to identify the hepatobiliary cysts with ovarian-type stroma and also included in the study (12 cases).
2 | www.ajsp.com
Volume 00, Number 00, ’’ 2017
All 36 cases that had undergone resection of the cyst and showed unequivocal ovarian stroma on microscopic examination were subjected to further detailed analysis. There was an average of 14.5 slides of the cyst examined per case (range, 4 to 45).
Clinicopathologic Parameters Clinical and demographic data, including age, sex, preoperative imaging findings, tumor recurrence, and follow-up period were obtained from the medical records. Cyst (tumor) size was obtained from the pathology reports, except in cases where the cyst was submitted in multiple specimens either obtained concurrently or in separate operations, the size was obtained from the most recent preoperative (magnetic resonance imaging [MRI] or computed tomography) radiology report. The presence and size of intracystic papillary/polypoid nodules was determined through gross and microscopic examination. The composition of such papillary/polypoid nodules (either complex epithelial proliferations or nonepithelial mural nodules comprising polypoid projections of ovarian stroma with or without daughter cysts, fibrosis, hemorrhage, inflammation, etc.) was also recorded. The most recent preoperative radiology report was reviewed for mention of intracystic papillary/mural nodules for comparison to pathologic findings. The amount of OS was estimated semi-quantitatively as focal (< 25% of the cyst wall/lining showing OS presence), moderate (25% to 50%), substantial ( > 50% to 75%) or diffuse ( > 75%). The lining epithelium was classified as either mucinous (Fig. 1) or biliary nonmucinous (Fig. 2) as done previously as a part of a different study.28 Briefly, if it resembled endocervical or gastric epithelium (or having the appearance that would have qualified as “pancreatic intraepithelial neoplasia” or “biliary intraepithelial neoplasia”),10,29,30 it was classified as “mucinous,” versus biliary nonmucinous when the epithelium resembled the bile ductular epithelium (with cuboidal cells and minimal cytoplasm).28,31,32 The degree of atypia (dysplasia) in the epithelium was graded both by the earlier 3-tiered grading system (low, intermediate, and high-grade) per WHO 2010 as well as by the recently modified 2-tiered grading system as low and high as has been adopted for intraepithelial neoplasia in the pancreas.8,10 Accordingly, if there was complex papillary configuration lined by severely atypical cells or prominent tufting or pseudostratification with cellular disorganization, pleomorphism, nuclear enlargement, irregular chromatin clumping, nucleolar prominence and nuclear contour irregularities, this was qualified as highgrade dysplasia (HGD). The amount of HGD/carcinoma in situ (CIS) was estimated semi-quantitatively using the same cutoffs applied for the ovarian stroma.
Comparison With Pancreatic MCNs
The findings in the study group were also compared with those in 178 pancreatic examples, some
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Am J Surg Pathol
Volume 00, Number 00, ’’ 2017
Hepatobiliary Mucinous Cystic Neoplasms
TABLE 1. General Characteristics of 36 Hepatobiliary MCNs Characteristic
n/N (%) Without Invasion With Invasion
No. cases 34/36 (94.4) Mean age (y) 50.4 Mean cyst (tumor) size (cm) 11.2 HGD/CIS 0 Mural nodule of any composition ( ≥ 1 cm) Prevalence 7/34 (20.6) Mean size (cm) 1.5 Intracystic papillary epithelial nodule ( ≥ 1 cm) Prevalence 0 Mean size (cm) NA Left lobe 23/31 (74.2) Central liver 1/31 (3.2) Right lobe 4/31 (12.9) Left lobe and extrahepatic bile duct 1/31 (3.2) Extrahepatic bile duct exclusively 2/31 (6.5) Diffuse OS prevalence 16/34 (47) Focal OS prevalence 3/34 (8.8) MCN in radiologic DDx 13/28 (46.4) Malignancy in radiologic DDx 4/28 (14.3)
2/36 (5.6) 61 17.5 2/2 (100) 2/2 (100) 2.2 2/2 (100) 2.2 0 0 1/1 (100) 0 0 1/2 (50) 0 1/2 (50) 2/2 (100)
DDx indicates differential diagnosis; NA, not available.
FIGURE 1. Tall columnar mucinous epithelium, as seen here, was seen at least focally in most of the cases. The ovarian stroma in this area is quite cellular (hematoxylin and eosin stain).
characteristics of which were already published as a different study.18
RESULTS General Characteristics General characteristics of the cases are summarized in Table 1. Defined by ovarian stroma, all 36 cases occurred in women (100%) and predominantly in the left
FIGURE 2. Nonmucinous cuboidal epithelium, as seen here, was seen at least focally in most cases and predominated in half. The ovarian stroma in this example is less cellular (hematoxylin and eosin stain). Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
lobe of the liver (72%). The mean age was 51 years old (range, 28 to 76 y). The mean tumor size was 11.6 cm (range, 5 to 23 cm). Twenty-nine (91%) were exclusively intrahepatic (left lobe: 23 [72%]; right lobe: 5 [16%]; central: 1 [3%]), 1 (3%) involved both the left lobe of the liver and the common hepatic duct, and 2 (6%) involved an extrahepatic bile duct exclusively.
Clinical Findings Most patients presented with abdominal pain (14/24; 58%); fewer presented with elevated liver enzymes, a mass or abdominal tenderness detected on routine physical examination, or an incidental mass on imaging (2 patients each; 8%); and 1 patient each (4%) presented with fever/ chills/jaundice/right upper quadrant soreness, right upper quadrant fullness and flank pain, nausea/vomiting, and one patient palpated an abdominal mass herself. The preoperative radiographic diagnosis/differential diagnosis was known in 30 of 36 cases; in 21 of these 30 cases, the preoperative radiology report was available for review and in the remaining 9 cases, the radiographic impression was gleaned from the clinical notes. The modality of preoperative imaging was known in 21 cases (11 magnetic resonance imaging and 10 computed tomography). The diagnosis of “biliary cystadenoma” or “cystadenocarcinoma” were at least suggested in the radiographic differential diagnosis in 14 of 30 cases; in an additional 11 of 30 cases, radiographic studies showed a cyst without mention of specific differential diagnosis. Two were thought to be abscesses (in one of these the possibility of an underlying cholangiocarcinoma was suggested) and in one case each the radiographic differential diagnosis mentioned metastasis, ruptured hepatic adenoma, or hemorrhagic bile duct hamartoma. Overall, 6 of 30 cases had possibility of malignancy mentioned in the www.ajsp.com
|3
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Quigley et al
Am J Surg Pathol
Volume 00, Number 00, ’’ 2017
radiographic differential diagnosis and only 2 of these proved to have carcinomatous transformation in the subsequent histologic analysis, both of which were minimally invasive (see below).
Original Interpretation and Diagnoses Of the 229 liver cyst cases systematically reviewed for this study, 8 of 32 cases (25%) that had originally been designated hepatobiliary cystadenoma/cystadenocarcinoma did not have OS and were reclassified as other (5 non-neoplastic, 2 intraductal papillary neoplasm of the bile duct, and 1 cystic cholangiocarcinoma). Of the 34 noninvasive mucinous cystic neoplasms, the original pathologic diagnoses rendered were hepatobiliary cystadenoma, mucinous cystadenoma, or biliary cystadenoma in 25 of 34 cases and mucinous cystic neoplasm in 9. The 2 cases with invasion were originally diagnosed as “mucinous cystadenocarcinoma arising in cystadenoma” and “hepatobiliary cystadenocarcinoma with microscopic foci of invasion.”
Pathologic Findings All cases by definition had at least some foci of OS. However, detectable OS was focal in 3 of 36 (8%), substantial in 16 of 36 (45%), and diffuse in 17 of 36 (47%). Three cases (8%) had focal epithelioid cells with larger cytoplasm suggestive of luteinization in the OS. The percentages of biliary nonmucinous epithelium in these cases was already documented as a part of a separate study.28 Twenty-nine of 36 cases had at least some areas with tall columnar mucinous epithelium (Fig. 1); this was predominant in 18 cases. The remaining 18 cases (50%) were lined predominantly by cuboidal nonmucinous epithelium (Fig. 2). Eight cases showed goblet cell (intestinal) metaplasia. All cases showed at least focal degenerative changes in the form of hemorrhage, calcification, necrosis, xanthogranulomatous inflammation, or stromal granulation tissue/abscess formation. Stromal inflammation was noted in 35/36 cases and was prominent in 4 cases. The inflammatory cells were predominantly lymphoplasmacytic in 26 (72%) accompanied by neutrophils in 56% (n = 20) and eosinophils in 8% (n = 3). In 9 cases (25%), stromal inflammatory cells were predominantly macrophages, because of intracystic hemorrhage and/or rupture, but also contained lymphoplasmacytic cells and neutrophils. In the 4 cases with prominent inflammation, the inflammation was composed mostly of lymphoplasmacytic cells and a substantial number of neutrophils, with eosinophils in 2 and macrophages in 1. Immunohistochemical stains performed on selected cases showed that the ovarian type stroma was positive for PR in 8 of 9, ER in 5 of 7 and inhibin in 2 of 6.
Degree of Dysplasia; Frequency of Carcinoma On the basis of the highest grade area, using the old 3-tiered grading system, 32 of 36 were classified as lowgrade dysplasia (Figs. 1, 2), 2 of 36 as intermediate, and 2 of 36 as HGD (Fig. 3). Along these lines, using the 2-tiered
4 | www.ajsp.com
FIGURE 3. Two cases with high-grade dysplasia growing in a complex papillary architecture (A, B) (hematoxylin and eosin stain; inset: higher magnification).
grading, only 2 cases qualified as “HGD” (CIS). Both cases with HGD/CIS had microscopic foci of invasion (measuring 7 and 8 mm) (2/36; 6%) (Fig. 4). No other case had invasion. The overall sizes of the MCNs with microscopic invasion were 17 and 18 cm. The HGD was substantial (50% to 75% of the lesion) in both cases. Papillary/polypoid nodules comprising complex epithelial proliferation (Fig. 3) (measuring 2.0 and 2.3 cm) within the cystic lesions were seen grossly or microscopically only in the 2 cases with HGD/CIS and microscopic foci of invasion (2/36; 6%). No case with low-grade or intermediategrade dysplasia had any complex papillary nodular epithelial proliferation. That said, grossly or microscopically recognizable “mural nodules” comprising either nodular protrusions of ovarian stroma with or without daughter cysts (Fig. 5) or combinations of fibrosis, inflammation, and hemorrhage were present in 12 of the 34 noninvasive cases (35%); 7 of these measured at least 1.0 cm (21%) and the Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Am J Surg Pathol
Volume 00, Number 00, ’’ 2017
Hepatobiliary Mucinous Cystic Neoplasms
FIGURE 5. Many of the low-grade MCNs had grossly or microscopically detectable papillary/polypoid mural nodules. In this example, the nodule is created by protrusion of mural daughter cysts. This nodule measures 6 mm across and 3 mm tall. None of the low-grade MCNs had any polypoid/papillary nodules composed of complex epithelial proliferation (hematoxylin and eosin stain).
Follow-up
FIGURE 4. Foci of carcinomatous change in 2 different MCNs (both hematoxylin and eosin stain) (A, B).
largest was 2.0 cm. Preoperative radiology reports were available for review in 21 cases, including one of the 2 with HGD/CIS and microscopic invasion. In the radiology reports, vague reference to nodularity was briefly mentioned in three of 21 cases (14%) without further description as to the number and size of nodules. Two of these proved to be low-grade MCNs. One with “nodular projections” was radiographically suspected to be a ruptured hepatic adenoma with hemorrhage and did not have any mural nodule on pathology; the other with “irregularity and nodularity of the septations” was suspected to be a “biliary cystadenoma or cystadenocarcinoma” on radiology and had a 0.8 cm mural nodule comprising fibrosis, inflammation, hemorrhage, and cholesterol clefts microscopically. For the case with HGD/CIS and microscopic invasion that had a 2.0 cm papillary epithelial nodule microscopically, the radiology report stated that there were “several areas of increased density of internal locules” and that “upon contrast administration, many of the septations enhance” and that “none of the more nodular looking components appears to enhance” (without any other reference to nodules). Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Follow-up information was available in 32 cases and in a median follow-up of 18.5 months since the patients’ first cyst resection for which the pathology slides were available for confirmation of diagnosis, 29 were alive with no evidence of disease, 2 patients had persistence/ recurrence of low-grade MCNs which were subsequently reresected (one after 7 mo and another after 15 y 11 mo), 1 patient was alive with reportedly “residual cysts” on MRI (these were not completely resected due to extent and location) and no patient died of disease. On follow-up, one invasive case had “residual cyst or hematoma” at 3.9 months, and the other was without disease at 36.6 months. The invasive case with 3.9 months of followup has been lost to follow-up for more than a decade. This case had persisted after initial unroofing performed 4 months before definitive resection. In many of these patients who underwent surgical resection of the MCN (16/28), initial noncurative procedures (such as drainage, marsupialization, fenestration, unroofing, cystojejunostomy, or other incomplete resection) were performed before the last operation. One of these patients who proved to have a low-grade MCN without invasion initially had a cystojejunostomy performed 23 years before the definitive resection (the pathologic diagnosis cannot be confirmed from the original operation in this patient). Another patient had persistence/recurrence of a hepatic low-grade MCN nearly 16 years after initial (probably incomplete) excision on which the pathology material from both operations has been reviewed and the diagnoses confirmed. An additional case persisted after known incomplete resection secondary to extent/location of the cysts. www.ajsp.com
|5
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Quigley et al
Am J Surg Pathol
Comparison With Pancreatic MCNs (With OS) When compared with pancreatic MCNs (with OS) (part of another study18), the hepatobiliary cases showed less frequent carcinomatous change. In hepatobiliary MCNs in this study, 6% had at least HGD/CIS (vs. 23.6% in pancreatic) and invasive carcinoma was also less frequent (6% vs. 16.3%).18 Similar to pancreatic MCNs, hepatobiliary MCNs occurred predominantly in women (all hepatic MCNs in this study were women, whereas in our pancreatic MCN group there were 2/178 men with proven OS). Hepatobiliary MCNs are much larger at the time of diagnosis compared with pancreatic counterparts (invasive cases: 17.5 vs. 9.4 cm in pancreatic; noninvasive cases: 11.2 vs. 5.4 cm in pancreatic). MCNs harboring carcinoma tend to be larger than those without (both carcinoma cases in this study were large, 17 and 18 cm, respectively, vs. average size of 11.2 cm in the noninvasive cases), similar to the findings in pancreatic ones (mean size in invasive cases, 9.4 vs. 5.4 cm in noninvasive cases). Also, MCNs with carcinoma tend to have intracystic papillary/polypoid epithelial nodules whereas this phenomenon was not observed in MCNs without carcinomatous transformation.
DISCUSSION Clinicopathologic characteristics of hepatobiliary MCNs defined by the presence of ovarian stroma (OS; as per the WHO 2010 criteria) have been difficult to glean from the literature owing to the lack of stringent requirement of OS for the diagnosis until recently. Most studies published under “hepatobiliary cystadenoma or cystadenocarcinoma” seemed to have included entities other than MCN (with OS) in the analysis. And indeed, 25% of the cases that had been classified as hepatobiliary cystadenoma/cystadenocarcinoma pathologically in our surgical database proved not to have OS by pathologic reexamination and thus not qualifying as MCN/OS by the current WHO criteria. Conversely, close to 80% of the MCN/OS cases had not been classified as MCN in the original pathology diagnosis. Thus, it is not surprising that one of the most comprehensive studies published on this topic recently23 includes 248 cases but about two thirds of them (164/248) did not have OS, and only 147 cases had mucinous epithelium which suggest that inhomogeneous disease were included. In addition, 13.3% were in men. The rate of malignant transformation in the literature ranged from 2% to 30%.2,23,31,33 This study demonstrates that, defined by the presence of OS, hepatobiliary MCNs have important distinctive characteristics substantially different than the impression in the literature. They occur (almost) exclusively in women. All patients in this study were women, but it can be expected that there will be some male patients as is the case with pancreatic MCNs (97% women, but we have seen rare male patients with OS).18 In addition, similar to their pancreatic counterparts, hepatobiliaryMCNs occur in perimenopausal women (mean age, 50 y)
6 | www.ajsp.com
Volume 00, Number 00, ’’ 2017
further supporting the role of altered hormonal milieu in their pathogenesis. Curiously, the left hepatic lobe, despite its smaller size, was the most frequent site of involvement (80% involved the left hepatic lobe; 72% were exclusively in the left lobe), in line with observations reported in previous studies (72% to 75% left lobe).31,33 In addition, only 2 of our 32 cases (6%) (where location was documented convincingly) was centered on an extrahepatic bile duct, whereas in another recent study 4 of 20 cases (20%) were centered on an extrahepatic bile duct31; this may be related to populational and consultation biases. According to the literature, some hepatobiliary MCNs that arise in the left hepatic lobe prolapse into the extrahepatic bile ducts34; one of our cases involved both the left liver lobe and an extrahepatic bile duct. The preferential localization in the left lobe is intriguing in terms of histopathogenesis of these tumors and brings up the question of whether “periductal fetal mesenchyme” which we believe OS recapitulates may be more prominent in the left lobe that is closer to the pancreatobiliary tract embryologically. Most importantly, once defined by the presence of OS, hepatobiliary MCNs (“hepatobiliary cystadenomas/ cystadenocarcinomas” with OS) rarely show carcinomatous transformation (6%) at the time of resection (see further discussions below), which is in full accordance with a recent paper by Zen et al33 which also used the OS requirement and showed carcinomatous changes in
Hepatobiliary Mucinous Cystic Neoplasms With Ovarian Type Stroma (So-Called “Hepatobiliary Cystadenoma/Cystadenocarcinoma”) Clinicopathologic Analysis of 36 Cases Illustrates Rarity of Carcinomatous Change Brian Quigley, MD,* Michelle D. Reid, MD,* Burcin Pehlivanoglu, MD,* Malcolm H. Squires, III, MD,† Shishir Maithel, MD,† Yue Xue, MD, PhD,* Choi Hyejeong, MD,‡ Gizem Akkas, MD,* Takashi Muraki, MD,* David A. Kooby, MD,† Juan M. Sarmiento, MD,† Ken Cardona, MD,† Aarti Sekhar, MD,§ Alyssa Krasinskas, MD,* and Volkan Adsay, MD8
Abstract: The literature is highly conflicting on hepatobiliary mucinous cystic neoplasms (MCNs), aka “hepatobiliary cystadenoma/ cystadenocarcinoma,” largely because ovarian stroma (OS) was not a requirement until WHO-2010 and is not widely applied even today. In this study, MCNs (with OS) accounted for 24 of 229 (11%) resected hepatic cysts in one institution. Eight of the 32 (25%) cysts that had been originally designated as hepatobiliary cystadenoma/ cystadenocarcinoma at the time of diagnosis proved not to have an OS during this review and were thus re-classified as non-MCN. In total, 36 MCNs (with OS) were analyzed—24 from the institutional files and 12 consultation cases. All were women. Mean age was 51 (28 to 76 y). Mean size was 11 cm (5 to 23 cm). Most (91%) were intrahepatic and in the left lobe (72%). Preoperative imaging mentioned “neoplasm” in 14 (47%) and carcinoma was a differential in 6 (19%) but only 2 proved to have carcinoma. Microscopically, only 47% demonstrated diffuse OS ( > 75% of the cyst wall/lining); OS was often focal. The cyst lining was often composed of non-mucinous biliary epithelium, and this was predominant in 50% of the cases. Degenerative changes of variable amount were seen in most cases. In situ and invasive carcinoma was seen in only 2 cases (6%), both with small invasion (7 and 8 mm). Five cases had persistence/ recurrence, 2 confirmed operatively (at 7 mo and 15 y). Of the 2 cases with carcinoma, one had “residual cyst or hematoma” by radiology at 4 months, and the other was without disease at 3 years. In conclusion, many cysts (25%) previously reported as hepatobiliary cystadenoma/cystadenocarcinoma are not MCNs. True MCNs are uncommon among resected hepatic cysts (11%), occur exclusively in females, are large, mostly intrahepatic and in the From the Departments of *Pathology; †Surgery; §Radiology, School of Medicine, Emory University, Atlanta, GA; ‡Department of Pathology, Ulsan University College of Medicine, Ulsan, Korea; and 8Department of Pathology, Medical College of Wisconsin Milwaukee, WI, USA. Conflicts of Interest and Source of Funding: The authors have disclosed that they have no significant relationships with, or financial interest in, any commercial companies pertaining to this article. Correspondence: Volkan Adsay, MD, Department of Pathology, Medical College of Wisconsin Milwaukee, WI 53226 (e-mail: [email protected]). Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Am J Surg Pathol
Volume 00, Number 00, ’’ 2017
left lobe (72%). Invasive carcinomas are small and uncommon (6%) compared with their pancreatic counterpart (16%). Recurrences are not uncommon following incomplete excision. Key Words: mucinous cystic neoplasm, hepatobiliary cystadenoma, hepatobiliary cystadenocarcinoma, liver cyst, hepatic cyst (Am J Surg Pathol 2017;00:000–000)
hen “hepatobiliary cystadenoma/cystadenocarcinoma” was first presented to the literature as an entity, the category was interpreted to embrace essentially any epitheliallined cystic lesion occurring in the liver and biliary tract.1–4 As more knowledge accumulated about the nature and classification of cystic lesions in the hepato-pancreatobiliary tract, the view on the criteria needed to make this diagnosis in the liver is now changing dramatically, many of them adapted from the pancreas. In the 1980s, essentially 2 types of cystic tumors had been characterized in the pancreas: mucinous and serous. Over the years, while the definitions of serous tumors remained unchanged,5 those for the “mucinous” category have undergone major modifications. First, cyst-forming intraductal tumors (lined by ductal/mucinous cells), namely, intraductal papillary mucinous neoplasms, intraductal oncocytic papillary neoplasms, and intraductal tubulopapillary neoplasms, were characterized.6–16 More recently simple cysts17 and other mimickers such as secondary cystic dilatations, congenital processes, and others have been delineated. The most important aspect of this paradigm shift in the pancreas that was pertinent to hepatic cysts was that, it has become clear that the presence of ovarian stroma delineates a pathologically and biologically distinct cyst type, and therefore, WHO established ovarian stroma (OS) as a requirement for the diagnosis of mucinous cystic neoplasm (MCN), both for the pancreas and liver. This stroma has all the histologic and immunophenotypic
W
www.ajsp.com
|1
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Quigley et al
Am J Surg Pathol
characteristics of ovarian cortical stroma, including the presence of luteal type cells. Defined by OS, in the pancreas MCN has very distinctive features including almost exclusive ( > 98%) occurrence in women and vast majority in the body/tail as a well-demarcated cystic mass,18 which is substantially different than the earlier impressions on this entity19–22 before the OS requirement. In fact, there are still publications without histologic verification that lead to a different clinical picture.23 Once defined by OS, resected pancreatic MCNs harbor invasive carcinoma in 16%, mostly of tubular type but occasionally of the sarcomatoid/undifferentiated type.18,24–27 The criteria established for the classification of cystic tumors in the pancreas is now also adopted to the hepatobiliary tract in which the nonspecific and tainted diagnosis of “hepatobiliary cystadenoma/cystadenocarcinoma” as an entity has been eliminated by the WHO-2010, and replaced by the more specific entities including MCN (defined by the presence of OS) as well as hepatobiliary counterparts of cystic intraductal tumors, namely intraductal papillary neoplasms, intraductal tubulopapillary neoplasms, and intraductal oncocytic papillary neoplasms. It is also being appreciated that not all biliary-lined cysts are MCNs, and that there is radiographic overlap with cystic bile duct hamartomas, congenital cysts, and Caroli disease. However, most of the current impressions in the literature about cystic hepatobiliary lesions are still based on the earlier criteria and literature of hepatobiliary cystadenoma/cystadenocarcinoma.23 The purpose of this study was to elucidate the relative frequency of hepatobiliary MCNs (as now defined by the WHO by the presence of OS) among hepatic cysts; determine its clinicopathologic characteristics and establish the true frequency of carcinomatous changes in this tumor type.
MATERIALS AND METHODS Appropriate institutional approvals were obtained for this retrospective review.
Case Selection All 24 tumors that had been designated as hepatic or biliary cystadenoma or cystadenocarcinoma or mucinous cystic neoplasm in the authors’ institutional files were retrieved and examined. Separately, 229 hepatic cystic lesions resected specifically in the authors’ institution between the years of 1994 and 2014 were reviewed specifically for the presence or absence of ovarian stroma. Only cases with unequivocal ovarian stroma —comprising densely packed spindle cells with elongated nuclei and scant cytoplasm underneath the epithelium—were included in the study. In cases where the ovarian stroma was focal or less cellular (eg, atrophic), its nature was confirmed with immunohistochemical staining for progesterone receptor (PR), estrogen receptor (ER) and inhibin. In addition, pathology consultation files of the authors were screened to identify the hepatobiliary cysts with ovarian-type stroma and also included in the study (12 cases).
2 | www.ajsp.com
Volume 00, Number 00, ’’ 2017
All 36 cases that had undergone resection of the cyst and showed unequivocal ovarian stroma on microscopic examination were subjected to further detailed analysis. There was an average of 14.5 slides of the cyst examined per case (range, 4 to 45).
Clinicopathologic Parameters Clinical and demographic data, including age, sex, preoperative imaging findings, tumor recurrence, and follow-up period were obtained from the medical records. Cyst (tumor) size was obtained from the pathology reports, except in cases where the cyst was submitted in multiple specimens either obtained concurrently or in separate operations, the size was obtained from the most recent preoperative (magnetic resonance imaging [MRI] or computed tomography) radiology report. The presence and size of intracystic papillary/polypoid nodules was determined through gross and microscopic examination. The composition of such papillary/polypoid nodules (either complex epithelial proliferations or nonepithelial mural nodules comprising polypoid projections of ovarian stroma with or without daughter cysts, fibrosis, hemorrhage, inflammation, etc.) was also recorded. The most recent preoperative radiology report was reviewed for mention of intracystic papillary/mural nodules for comparison to pathologic findings. The amount of OS was estimated semi-quantitatively as focal (< 25% of the cyst wall/lining showing OS presence), moderate (25% to 50%), substantial ( > 50% to 75%) or diffuse ( > 75%). The lining epithelium was classified as either mucinous (Fig. 1) or biliary nonmucinous (Fig. 2) as done previously as a part of a different study.28 Briefly, if it resembled endocervical or gastric epithelium (or having the appearance that would have qualified as “pancreatic intraepithelial neoplasia” or “biliary intraepithelial neoplasia”),10,29,30 it was classified as “mucinous,” versus biliary nonmucinous when the epithelium resembled the bile ductular epithelium (with cuboidal cells and minimal cytoplasm).28,31,32 The degree of atypia (dysplasia) in the epithelium was graded both by the earlier 3-tiered grading system (low, intermediate, and high-grade) per WHO 2010 as well as by the recently modified 2-tiered grading system as low and high as has been adopted for intraepithelial neoplasia in the pancreas.8,10 Accordingly, if there was complex papillary configuration lined by severely atypical cells or prominent tufting or pseudostratification with cellular disorganization, pleomorphism, nuclear enlargement, irregular chromatin clumping, nucleolar prominence and nuclear contour irregularities, this was qualified as highgrade dysplasia (HGD). The amount of HGD/carcinoma in situ (CIS) was estimated semi-quantitatively using the same cutoffs applied for the ovarian stroma.
Comparison With Pancreatic MCNs
The findings in the study group were also compared with those in 178 pancreatic examples, some
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Am J Surg Pathol
Volume 00, Number 00, ’’ 2017
Hepatobiliary Mucinous Cystic Neoplasms
TABLE 1. General Characteristics of 36 Hepatobiliary MCNs Characteristic
n/N (%) Without Invasion With Invasion
No. cases 34/36 (94.4) Mean age (y) 50.4 Mean cyst (tumor) size (cm) 11.2 HGD/CIS 0 Mural nodule of any composition ( ≥ 1 cm) Prevalence 7/34 (20.6) Mean size (cm) 1.5 Intracystic papillary epithelial nodule ( ≥ 1 cm) Prevalence 0 Mean size (cm) NA Left lobe 23/31 (74.2) Central liver 1/31 (3.2) Right lobe 4/31 (12.9) Left lobe and extrahepatic bile duct 1/31 (3.2) Extrahepatic bile duct exclusively 2/31 (6.5) Diffuse OS prevalence 16/34 (47) Focal OS prevalence 3/34 (8.8) MCN in radiologic DDx 13/28 (46.4) Malignancy in radiologic DDx 4/28 (14.3)
2/36 (5.6) 61 17.5 2/2 (100) 2/2 (100) 2.2 2/2 (100) 2.2 0 0 1/1 (100) 0 0 1/2 (50) 0 1/2 (50) 2/2 (100)
DDx indicates differential diagnosis; NA, not available.
FIGURE 1. Tall columnar mucinous epithelium, as seen here, was seen at least focally in most of the cases. The ovarian stroma in this area is quite cellular (hematoxylin and eosin stain).
characteristics of which were already published as a different study.18
RESULTS General Characteristics General characteristics of the cases are summarized in Table 1. Defined by ovarian stroma, all 36 cases occurred in women (100%) and predominantly in the left
FIGURE 2. Nonmucinous cuboidal epithelium, as seen here, was seen at least focally in most cases and predominated in half. The ovarian stroma in this example is less cellular (hematoxylin and eosin stain). Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
lobe of the liver (72%). The mean age was 51 years old (range, 28 to 76 y). The mean tumor size was 11.6 cm (range, 5 to 23 cm). Twenty-nine (91%) were exclusively intrahepatic (left lobe: 23 [72%]; right lobe: 5 [16%]; central: 1 [3%]), 1 (3%) involved both the left lobe of the liver and the common hepatic duct, and 2 (6%) involved an extrahepatic bile duct exclusively.
Clinical Findings Most patients presented with abdominal pain (14/24; 58%); fewer presented with elevated liver enzymes, a mass or abdominal tenderness detected on routine physical examination, or an incidental mass on imaging (2 patients each; 8%); and 1 patient each (4%) presented with fever/ chills/jaundice/right upper quadrant soreness, right upper quadrant fullness and flank pain, nausea/vomiting, and one patient palpated an abdominal mass herself. The preoperative radiographic diagnosis/differential diagnosis was known in 30 of 36 cases; in 21 of these 30 cases, the preoperative radiology report was available for review and in the remaining 9 cases, the radiographic impression was gleaned from the clinical notes. The modality of preoperative imaging was known in 21 cases (11 magnetic resonance imaging and 10 computed tomography). The diagnosis of “biliary cystadenoma” or “cystadenocarcinoma” were at least suggested in the radiographic differential diagnosis in 14 of 30 cases; in an additional 11 of 30 cases, radiographic studies showed a cyst without mention of specific differential diagnosis. Two were thought to be abscesses (in one of these the possibility of an underlying cholangiocarcinoma was suggested) and in one case each the radiographic differential diagnosis mentioned metastasis, ruptured hepatic adenoma, or hemorrhagic bile duct hamartoma. Overall, 6 of 30 cases had possibility of malignancy mentioned in the www.ajsp.com
|3
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Quigley et al
Am J Surg Pathol
Volume 00, Number 00, ’’ 2017
radiographic differential diagnosis and only 2 of these proved to have carcinomatous transformation in the subsequent histologic analysis, both of which were minimally invasive (see below).
Original Interpretation and Diagnoses Of the 229 liver cyst cases systematically reviewed for this study, 8 of 32 cases (25%) that had originally been designated hepatobiliary cystadenoma/cystadenocarcinoma did not have OS and were reclassified as other (5 non-neoplastic, 2 intraductal papillary neoplasm of the bile duct, and 1 cystic cholangiocarcinoma). Of the 34 noninvasive mucinous cystic neoplasms, the original pathologic diagnoses rendered were hepatobiliary cystadenoma, mucinous cystadenoma, or biliary cystadenoma in 25 of 34 cases and mucinous cystic neoplasm in 9. The 2 cases with invasion were originally diagnosed as “mucinous cystadenocarcinoma arising in cystadenoma” and “hepatobiliary cystadenocarcinoma with microscopic foci of invasion.”
Pathologic Findings All cases by definition had at least some foci of OS. However, detectable OS was focal in 3 of 36 (8%), substantial in 16 of 36 (45%), and diffuse in 17 of 36 (47%). Three cases (8%) had focal epithelioid cells with larger cytoplasm suggestive of luteinization in the OS. The percentages of biliary nonmucinous epithelium in these cases was already documented as a part of a separate study.28 Twenty-nine of 36 cases had at least some areas with tall columnar mucinous epithelium (Fig. 1); this was predominant in 18 cases. The remaining 18 cases (50%) were lined predominantly by cuboidal nonmucinous epithelium (Fig. 2). Eight cases showed goblet cell (intestinal) metaplasia. All cases showed at least focal degenerative changes in the form of hemorrhage, calcification, necrosis, xanthogranulomatous inflammation, or stromal granulation tissue/abscess formation. Stromal inflammation was noted in 35/36 cases and was prominent in 4 cases. The inflammatory cells were predominantly lymphoplasmacytic in 26 (72%) accompanied by neutrophils in 56% (n = 20) and eosinophils in 8% (n = 3). In 9 cases (25%), stromal inflammatory cells were predominantly macrophages, because of intracystic hemorrhage and/or rupture, but also contained lymphoplasmacytic cells and neutrophils. In the 4 cases with prominent inflammation, the inflammation was composed mostly of lymphoplasmacytic cells and a substantial number of neutrophils, with eosinophils in 2 and macrophages in 1. Immunohistochemical stains performed on selected cases showed that the ovarian type stroma was positive for PR in 8 of 9, ER in 5 of 7 and inhibin in 2 of 6.
Degree of Dysplasia; Frequency of Carcinoma On the basis of the highest grade area, using the old 3-tiered grading system, 32 of 36 were classified as lowgrade dysplasia (Figs. 1, 2), 2 of 36 as intermediate, and 2 of 36 as HGD (Fig. 3). Along these lines, using the 2-tiered
4 | www.ajsp.com
FIGURE 3. Two cases with high-grade dysplasia growing in a complex papillary architecture (A, B) (hematoxylin and eosin stain; inset: higher magnification).
grading, only 2 cases qualified as “HGD” (CIS). Both cases with HGD/CIS had microscopic foci of invasion (measuring 7 and 8 mm) (2/36; 6%) (Fig. 4). No other case had invasion. The overall sizes of the MCNs with microscopic invasion were 17 and 18 cm. The HGD was substantial (50% to 75% of the lesion) in both cases. Papillary/polypoid nodules comprising complex epithelial proliferation (Fig. 3) (measuring 2.0 and 2.3 cm) within the cystic lesions were seen grossly or microscopically only in the 2 cases with HGD/CIS and microscopic foci of invasion (2/36; 6%). No case with low-grade or intermediategrade dysplasia had any complex papillary nodular epithelial proliferation. That said, grossly or microscopically recognizable “mural nodules” comprising either nodular protrusions of ovarian stroma with or without daughter cysts (Fig. 5) or combinations of fibrosis, inflammation, and hemorrhage were present in 12 of the 34 noninvasive cases (35%); 7 of these measured at least 1.0 cm (21%) and the Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Am J Surg Pathol
Volume 00, Number 00, ’’ 2017
Hepatobiliary Mucinous Cystic Neoplasms
FIGURE 5. Many of the low-grade MCNs had grossly or microscopically detectable papillary/polypoid mural nodules. In this example, the nodule is created by protrusion of mural daughter cysts. This nodule measures 6 mm across and 3 mm tall. None of the low-grade MCNs had any polypoid/papillary nodules composed of complex epithelial proliferation (hematoxylin and eosin stain).
Follow-up
FIGURE 4. Foci of carcinomatous change in 2 different MCNs (both hematoxylin and eosin stain) (A, B).
largest was 2.0 cm. Preoperative radiology reports were available for review in 21 cases, including one of the 2 with HGD/CIS and microscopic invasion. In the radiology reports, vague reference to nodularity was briefly mentioned in three of 21 cases (14%) without further description as to the number and size of nodules. Two of these proved to be low-grade MCNs. One with “nodular projections” was radiographically suspected to be a ruptured hepatic adenoma with hemorrhage and did not have any mural nodule on pathology; the other with “irregularity and nodularity of the septations” was suspected to be a “biliary cystadenoma or cystadenocarcinoma” on radiology and had a 0.8 cm mural nodule comprising fibrosis, inflammation, hemorrhage, and cholesterol clefts microscopically. For the case with HGD/CIS and microscopic invasion that had a 2.0 cm papillary epithelial nodule microscopically, the radiology report stated that there were “several areas of increased density of internal locules” and that “upon contrast administration, many of the septations enhance” and that “none of the more nodular looking components appears to enhance” (without any other reference to nodules). Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Follow-up information was available in 32 cases and in a median follow-up of 18.5 months since the patients’ first cyst resection for which the pathology slides were available for confirmation of diagnosis, 29 were alive with no evidence of disease, 2 patients had persistence/ recurrence of low-grade MCNs which were subsequently reresected (one after 7 mo and another after 15 y 11 mo), 1 patient was alive with reportedly “residual cysts” on MRI (these were not completely resected due to extent and location) and no patient died of disease. On follow-up, one invasive case had “residual cyst or hematoma” at 3.9 months, and the other was without disease at 36.6 months. The invasive case with 3.9 months of followup has been lost to follow-up for more than a decade. This case had persisted after initial unroofing performed 4 months before definitive resection. In many of these patients who underwent surgical resection of the MCN (16/28), initial noncurative procedures (such as drainage, marsupialization, fenestration, unroofing, cystojejunostomy, or other incomplete resection) were performed before the last operation. One of these patients who proved to have a low-grade MCN without invasion initially had a cystojejunostomy performed 23 years before the definitive resection (the pathologic diagnosis cannot be confirmed from the original operation in this patient). Another patient had persistence/recurrence of a hepatic low-grade MCN nearly 16 years after initial (probably incomplete) excision on which the pathology material from both operations has been reviewed and the diagnoses confirmed. An additional case persisted after known incomplete resection secondary to extent/location of the cysts. www.ajsp.com
|5
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Quigley et al
Am J Surg Pathol
Comparison With Pancreatic MCNs (With OS) When compared with pancreatic MCNs (with OS) (part of another study18), the hepatobiliary cases showed less frequent carcinomatous change. In hepatobiliary MCNs in this study, 6% had at least HGD/CIS (vs. 23.6% in pancreatic) and invasive carcinoma was also less frequent (6% vs. 16.3%).18 Similar to pancreatic MCNs, hepatobiliary MCNs occurred predominantly in women (all hepatic MCNs in this study were women, whereas in our pancreatic MCN group there were 2/178 men with proven OS). Hepatobiliary MCNs are much larger at the time of diagnosis compared with pancreatic counterparts (invasive cases: 17.5 vs. 9.4 cm in pancreatic; noninvasive cases: 11.2 vs. 5.4 cm in pancreatic). MCNs harboring carcinoma tend to be larger than those without (both carcinoma cases in this study were large, 17 and 18 cm, respectively, vs. average size of 11.2 cm in the noninvasive cases), similar to the findings in pancreatic ones (mean size in invasive cases, 9.4 vs. 5.4 cm in noninvasive cases). Also, MCNs with carcinoma tend to have intracystic papillary/polypoid epithelial nodules whereas this phenomenon was not observed in MCNs without carcinomatous transformation.
DISCUSSION Clinicopathologic characteristics of hepatobiliary MCNs defined by the presence of ovarian stroma (OS; as per the WHO 2010 criteria) have been difficult to glean from the literature owing to the lack of stringent requirement of OS for the diagnosis until recently. Most studies published under “hepatobiliary cystadenoma or cystadenocarcinoma” seemed to have included entities other than MCN (with OS) in the analysis. And indeed, 25% of the cases that had been classified as hepatobiliary cystadenoma/cystadenocarcinoma pathologically in our surgical database proved not to have OS by pathologic reexamination and thus not qualifying as MCN/OS by the current WHO criteria. Conversely, close to 80% of the MCN/OS cases had not been classified as MCN in the original pathology diagnosis. Thus, it is not surprising that one of the most comprehensive studies published on this topic recently23 includes 248 cases but about two thirds of them (164/248) did not have OS, and only 147 cases had mucinous epithelium which suggest that inhomogeneous disease were included. In addition, 13.3% were in men. The rate of malignant transformation in the literature ranged from 2% to 30%.2,23,31,33 This study demonstrates that, defined by the presence of OS, hepatobiliary MCNs have important distinctive characteristics substantially different than the impression in the literature. They occur (almost) exclusively in women. All patients in this study were women, but it can be expected that there will be some male patients as is the case with pancreatic MCNs (97% women, but we have seen rare male patients with OS).18 In addition, similar to their pancreatic counterparts, hepatobiliaryMCNs occur in perimenopausal women (mean age, 50 y)
6 | www.ajsp.com
Volume 00, Number 00, ’’ 2017
further supporting the role of altered hormonal milieu in their pathogenesis. Curiously, the left hepatic lobe, despite its smaller size, was the most frequent site of involvement (80% involved the left hepatic lobe; 72% were exclusively in the left lobe), in line with observations reported in previous studies (72% to 75% left lobe).31,33 In addition, only 2 of our 32 cases (6%) (where location was documented convincingly) was centered on an extrahepatic bile duct, whereas in another recent study 4 of 20 cases (20%) were centered on an extrahepatic bile duct31; this may be related to populational and consultation biases. According to the literature, some hepatobiliary MCNs that arise in the left hepatic lobe prolapse into the extrahepatic bile ducts34; one of our cases involved both the left liver lobe and an extrahepatic bile duct. The preferential localization in the left lobe is intriguing in terms of histopathogenesis of these tumors and brings up the question of whether “periductal fetal mesenchyme” which we believe OS recapitulates may be more prominent in the left lobe that is closer to the pancreatobiliary tract embryologically. Most importantly, once defined by the presence of OS, hepatobiliary MCNs (“hepatobiliary cystadenomas/ cystadenocarcinomas” with OS) rarely show carcinomatous transformation (6%) at the time of resection (see further discussions below), which is in full accordance with a recent paper by Zen et al33 which also used the OS requirement and showed carcinomatous changes in
