Download - Adsay icpn 2012

8/16/2019 Adsay icpn 2012

1/23

Intracholecystic Papillary-Tubular Neoplasms (ICPN)of the Gallbladder (Neoplastic Polyps, Adenomas,

and Papillary Neoplasms That Are Z1.0 cm)Clinicopathologic and Immunohistochemical

Analysis of 123 Cases

Volkan Adsay, MD,* Kee-Taek Jang, MD,w Juan Carlos Roa, MD,z Nevra Dursun, MD,y Nobuyuki Ohike, MD,8 Pelin Bagci, MD,z Olca Basturk, MD,# Sudeshna Bandyopadhyay, MD,**

Jeanette D. Cheng, MD,ww Juan M. Sarmiento, MD,zz Oscar Tapia Escalona, MD,z

Michael Goodman, MD,yy So Yeon Kong, MPH,yy and Paul Terry, PhD, MPH 88

Abstract: The literature on the clinicopathologic characteristics

of tumoral intraepithelial neoplasms (neoplastic polyps) of the

gallbladder (GB) is fairly limited, due in part to the variability in

definition and terminology. Most reported adenomas (pyloric

gland type and others) were microscopic and thus regarded as

clinically inconsequential, whereas papillary in situ carcinomas

have been largely considered a type of invasive adenocarcinoma

under the heading of “papillary adenocarcinomas.” In this

study, 123 GB cases that have a well-defined exophytic pre-

invasive neoplasm measuring Z1 cm were analyzed. The pa-

tients were predominantly female (F/M = 2:1) with a mean age

of 61 y and a median tumor size of 2.2 cm. Half of the patientspresented with pain, and in the other half the neoplasm was

detected incidentally. Other neoplasms, most being gastro-

intestinal tract malignancies, were present in 22% of cases.

Gallstones were identified in only 20% of cases. Radiologically,

almost half were diagnosed as “cancer,” roughly half with pol-

ypoid tumor, and in 10% the lesion was missed. Pathologic

findings: (1) The predominant configuration was papillary in

43%, tubulopapillary in 31%, tubular in 26%. (2) Each case was

assigned a final lineage type on the basis of the predominant

pattern (> 75% of the lesion) on morphology, and supported

with specific immunohistochemical cell lineage markers. The

predominant cell lineage could be identified as biliary in 50%

(66% of which were MUC1+), gastric foveolar in 16% (all were

MUC5AC+), gastric pyloric in 20% (92% MUC6+), intestinal

in 8% (100% CK20+; 75% CDX2+; 50%, MUC2+), and on-

cocytic in 6% (17% HepPar+ and 17% MUC6+); however,

90% of cases had some amount of secondary or unclassifiable

pattern and hybrid immunophenotypes. (3) Of the cases that

would have qualified as “pyloric gland adenoma,” 21/24 (88%)

had at least focal high-grade dysplasia and 18% had associated

invasive carcinoma. Conversely, 8 of 47 “papillary adenocar-

cinoma”-type cases displayed some foci of low-grade dysplasia,

and 15/47 (32%) had no identifiable invasion. (4) Overall,

55% of the cases had an associated invasive carcinoma (pan-

creatobiliary type, 58; others, 10). Factors associated signi-

ficantly with invasion were the extent of high-grade dysplasia,

cell type (biliary or foveolar), and papilla formation. Amongsystematically analyzed invasive carcinomas, tumoral intra-

epithelial neoplasia was detected in 6.4% (39/606). (5) The 3-

year actuarial survival was 90% for cases without invasion and

60% for those associated with invasion. In contrast, those as-

sociated with invasion had a far better clinical outcome com-

pared with pancreatobiliary-type GB carcinomas (3-yr survival,

27%), and this survival advantage persisted even with stage-

matched comparison. Death occurred in long-term follow-up

even in a few noninvasive cases (4/55; median 73.5 mo) em-

phasizing the importance of long-term follow-up. In conclusion,

tumoral preinvasive neoplasms (Z1 cm) in the GB are analo-

gous to their pancreatic and biliary counterparts (biliary

From the Departments of *Pathology; zzSurgery; wwPiedmont Hospital,Emory University School of Medicine; yyDepartment of Epidemio-logy, Emory University School of Public Health, Atlanta, GA;#Memorial Sloan-Kettering Cancer Center, New York, NY; **TheKarmanos Cancer Institute and Wayne State University, Detroit,MI; 88Department of Public Health, College of Education, Health &Human Sciences, The University of Tennessee of Knoxville, TN;wDepartment of Pathology, Samsung Medical Center, Sungkyunk-wan University School of Medicine, Seoul, Korea; zDepartment of Pathology, Frontera University School of Medicine, Temuco, Chile;yDepartment of Pathology, Istanbul Education and Research Hos-pital, Istanbul, Turkey; 8First Department of Pathology, Showa

University School of Medicine, Tokyo, Japan; and zDepartment of Pathology, Rize University, School of Medicine, Rize, Turkey.V.A. and K-T.J. contributed equally.Presented in part at the annual meeting of the United States and

Canadian Academy of Pathology in Washington, DC, March 2010,and San Antonio, TX, March 2011.

Conflicts of Interest and Source of Funding: Supported in part byFondecyt Grant #1090171, Chile, and in part by the Georgia CancerCoalition Distinguished Cancer Clinicians and Scientists Program,GA. The authors have disclosed that they have no significant rela-tionships with, or financial interest in, any commercial companiespertaining to this article.

Correspondence: Volkan Adsay, MD, Department of Pathology andLaboratory Medicine, Emory University Hospital, 1364 CliftonRoad NE, Atlanta, GA 30322 (e-mail: [email protected]).

Copyright r 2012 by Lippincott Williams & Wilkins

ORIGINAL ARTICLE

Am J Surg Pathol Volume 36, Number 9, September 2012 www.ajsp.com | 1279

mailto:[email protected]:[email protected]

8/16/2019 Adsay icpn 2012

2/23

intraductal papillary neoplasms, pancreatic intraductal papillary

mucinous neoplasms, and intraductal tubulopapillary neo-

plasms). They show variable cellular lineages, a spectrum of

dysplasia, and a mixture of papillary or tubular growth patterns,

often with significant overlap, warranting their classification

under 1 unified parallel category, intracholecystic papillary-

tubular neoplasm. Intracholecystic papillary-tubular neoplasmsare relatively indolent neoplasia with significantly better prog-

nosis compared with pancreatobiliary-type GB carcinomas. In

contrast, even seemingly innocuous examples such as those re-

ferred to as “pyloric gland adenomas” can progress to carci-

noma and be associated with invasion and fatal outcome.

Key Words: gallbladder, adenoma, carcinoma, in situ, pre-

invasive neoplasm, papillary, pyloric, tubular, intestinal,

oncocytic, biliary

(Am J Surg Pathol 2012;36:1279–1301)

In the pancreatobiliary tract, tumors composed of pre-invasive neoplastic (dysplastic) cells that form clinicallydetectable (Z1.0 cm) masses are now classified under aunified category of intraductal papillary neoplasms(IPNs) in the bile ducts1 and as intraductal papillarymucinous neoplasms (IPMNs) or intraductal tubulopa-pillary neoplasms (ITPNs) in the pancreas2 (Fig. 1). It isnow well established that these preinvasive neoplasms(what we term tumoral intraepithelial neoplasms) repre-sent an “adenoma-carcinoma” sequence,3–18 and thattheir clinicopathologic, immunophenotypic, and molec-ular characteristics as well as biological behavior aredifferent from the nontumoral (flat)-type preinvasive ne-

oplasms of the respective organs.5,8,10,14,15,19 At the sametime, they are also distinct from the conventional invasivecancers of these sites for which they are often mistakenbecause of their mass-forming nature.2,10,20–34

In the pancreas, in which such tumors are bestcharacterized, the category of IPMN was created to en-compass a wide spectrum of lesions ranging from in-nocuous cysts lined by gastric foveolar epithelium withoutatypia (previously called “hyperplasia” by the JapanesePathology Society) and those that resemble villous ad-enomas, associated with extensive invasive carcinoma of the mucinous type [previously called “papillary mucinouscarcinoma” by World Health Organization (WHO)],35 to

those that have complex papillary architecture associatedwith invasive carcinoma of the pancreatobiliary type, whichused to be called “papillary adenocarcinomas.”26–28 Therare pyloric gland adenoma type lesions36,37 are now alsoregarded as part of the IPMN category.2 More recently,nonmucinous examples of tumoral intraepithelial neoplasiathat occur in the pancreas have also been characterized andhave been recognized by the WHO as ITPNs.2,28,38–41

Recognition of pancreatic IPMNs has led tothe reappraisal of preinvasive lesions in the biliarytract.20,22,42 Many authors originally adopted the term“biliary IPMN,”43–49 a category that encompasses tubu-lar, papillary, and villous preinvasive neoplasms including

papillomatosis.20,22,31,42–45,50–53 However, because mucinproduction is much less evident in these biliary examples,these are now unified under the heading of intraductalpapillary neoplasm (IPN) in the 2010 WHO classificationfor both intrahepatic and extrahepatic lesions.1 Likewise,we recently showed that in the ampulla of Vater, there

is occurrence of analogous tumoral intraepithelial neo-plasms with distinctive properties. Accordingly, we pro-posed the term intra-ampullary papillary-tubular neoplasms(IAPNs).54

A similar spectrum of lesions also exists but has notyet been fully characterized in the gallbladder (GB). Aplethora of names have been used including “pyloricgland adenoma,” “papillary adenoma,” “tubulopapillaryadenoma,” “intestinal adenoma,” “biliary adenoma,”“transitional adenoma,” “papillary neoplasm,” “papillarycarcinoma,” “intracystic papillary neoplasm,”1,15,55–77

and others, for many of which only a handful of caseshave been reported in the literature.78–87 It should be

noted here that some of these names are based on thegrowth pattern, some on cell lineage, and others on thedegree of neoplastic change, rendering this terminologydifficult to use, especially considering the frequency of overlap and occurrence of all 3 parameters in a given case.The most recent WHO classification attempted to addressthis issue by creating 2 generic categories, “adenoma”versus “intracystic papillary neoplasm,” each with severalsubsets. However, no criteria were provided as to howmuch papilla formation would qualify a lesion as intra-cystic papillary neoplasm rather than adenoma, or howmuch high-grade dysplasia (HGD) is allowable in the“adenoma” category.

FIGURE 1. Terminologic and conceptual analogy betweentumoral intraepithelial neoplasms occurring in the ampullo-pancreatobiliary tract. IPMN and ITPN of the pancreas and IPN

of the bile ducts are recognized by the WHO. IAPN is thecategory designation we recently proposed for those occur-ring in the ampulla. These show numerous similarities to“adenomas” and “intracystic papillary neoplasms” of the GB,and thus warranting these lesions to be unified under 1heading for which we propose the term ICPN.

Adsay et al Am J Surg Pathol Volume 36, Number 9, September 2012

1280 | www.ajsp.com r 2012 Lippincott Williams & Wilkins

8/16/2019 Adsay icpn 2012

3/23

Adenoma is, in fact, one of the most commonlyused terms in the literature for these lesions but refermostly, and sometimes exclusively, to “pyloric gland ad-enomas,”55,65,68,73,88,89 whereas other authors also includepapillary lesions or intestinal-type neoplasms15,33,58,67 inthe adenoma category. Furthermore, in the 3 largest

published series on “adenomas,” the mean “tumor” sizewas 7.2, 8.2, and 9.8 mm and included lesions as small as0.5, 1, and 3 mm,65,73,88 most of which may representpolypoid metaplasia. This created the impression thatadenomas are invariably benign33,88,90,91; although withcareful dissection of the literature, it becomes clear thatthe ones >1 cm are commonly associated with carcino-ma.13,15,59,68,73,74,90,92–100 At the other end of the spec-trum, papillary preinvasive neoplasms of the GB are alsotumoral intraepithelial neoplasms, but have mostly beenexcluded from the analysis of preinvasive neoplasms be-cause they are commonly associated with invasive carci-noma. They have typically been studied along with

conventional (pancreatobiliary-type) GB cancers as their“papillary adenocarcinoma” type, although it was re-cently shown by Albores-Saavedra et al57 that non-invasive cases, not surprisingly, may have a very goodprognosis. Further, the reported percentage of invasiveGB carcinomas, which arise from the “adenoma-carcinoma” sequence as opposed to the (flat) dysplasia-carcinoma sequence11,66,73,89,101–103 ranges widely, from5% to 23%.13,15,66,68,73 Thus, because of the highly vari-able definitions, along with the relative rarity of theseneoplasms, it has been difficult to determine the frequencyand clinicopathologic characteristics and, more im-portantly, their association with invasive carcinoma andbehavior.

The aim of this study is to analyze neoplastic pol-ypoid “adenomatous” and papillary preinvasive neoplasia(tumoral intraepithelial neoplasia) of the GB through theperspective of recently developed concepts and criteria forelsewhere in the pancreatobiliary tract (Fig. 1). Accord-ingly, we performed a detailed clinicopathologic analysisof 123 neoplasms Z1 cm in our surgical database, whichwe propose to classify under the conceptual category of intracholecystic papillary-tubular neoplasm (ICPN).

MATERIALS AND METHODSThe studies were conducted in accordance with the

Institutional Review Board requirements.

Case SelectionFrom the authors’ institutional and consultation

surgical pathology databases, all the cases carrying thediagnosis of GB “polyp,” “adenoma,” “neoplasm,” and“papillary” were retrieved. In addition, 3265 consecutiveroutine cholecystectomies performed in the authors’ in-stitutions (removed for gallstones and cholecystitis) and606 consecutive invasive GB carcinoma cases, includingearly GB carcinomas,73,104 were analyzed systematicallyto determine the frequency of these lesions. Cases thatfulfilled the following criteria were designated as ICPNand included in this study.

Inclusion CriteriaExophytic (papillary or polypoid) intramucosal GB

masses that measure Z1.0 cm and are composed of pre-invasive neoplastic (dysplastic) cells forming a compactlesion distinct from the neighboring mucosa (Table 1)(Figs. 2, 3) are included.

The criterion of Z1.0 cm is the same as that usedelsewhere in the pancreatobiliary tract to distinguish thesetumoral intraepithelial neoplasms (pancreatic IPMN,pancreatic ITPN, and the IPN of the extrahepatic andintrahepatic bile duct) from their nontumoral counter-parts (PanINs and BilINs).27 Furthermore, this Z1.0 cmmeasurement was also adopted as the specific definitionof these entities by the recent WHO classification.2,27

Lesions

8/16/2019 Adsay icpn 2012

4/23

Demographic and Clinical DataInformation on the patients’ gender, age, and the

clinical outcome was obtained from patients’ charts orfrom their primary physicians. Detailed clinical in-formation could be obtained on 62 of the patients. Forsome cases, follow-up information was obtained from theSurveillance Epidemiology End Results database.

Histomorphologic Analysis

Growth PatternsThe preinvasive lesions were evaluated for the

amount of papillary or tubular growth. For the purposes

of this study, each case was assigned to 1 of 3 groups: (a)papillary if the lesion had >75% papillary or villousgrowth; (b) tubular if >75% tubular growth; or (c) tu-bulopapillary if the secondary pattern constituted >25%of the preinvasive lesion.

DysplasiaDysplasia was graded as low or high33 using the

same cytologic criteria as those for flat lesions of the GB.This was preferred to the 3-tiered system that is currentlyin use elsewhere in the pancreatobiliary tract, because it isbecoming increasingly clear that the pancreatobiliarycases cluster into 2 groups, HGD/carcinoma in situ (CIS)versus non-HGD2,115–118 from both the biological andmanagement standpoints. Moreover, in all likelihood, theclassification will eventually revert to a 2-tiered system asin other organs such as the cervix and bladder.119–121 The

amount of HGD was scored as follows: focal, if it was75% of the lesion.

TABLE 1. Definition of ICPN

A GB neoplasm that is:IntramucosalPreinvasive neoplastic (dysplastic)Mass forming; exophytic (papillary or polypoid)Z1.0cm

CompactDistinct from the neighboring mucosa

FIGURE 2. ICPNs are characterized by a distinct polypoid mass protruding into the lumen (A). Some examples, especially thosewith tubular architecture, tend to be more lobulated and attached to the mucosa with a thin stalk (B). ICPNs with more papillaryconfiguration tend to have a more pedunculated growth (C, D).



8/16/2019 Adsay icpn 2012

5/23

Cell Lineage MorphologyCell lineage was determined on the basis of mor-

phologic criteria established for pancreatic IPMNs10,23,25

and also recently for intra-ampullary neoplasms.54 Thelesions were classified as intestinal (Fig. 4) if there wasclose resemblance to colonic adenomas or intestinal-typeIPMNs. Those that bear similarity to gastric foveolarmucosa, such as gastric foveolar IPMNs or gastric fo-veolar adenomas,122,123 were classified as gastric foveolar(Fig. 5). If the lesion looked like gastric pyloric glandadenomas,124 it was classified as gastric pyloric (Fig. 6).Those resembling intraductal oncocytic papillary neo-

plasms21,23

(oncocytic IPMNs) were regarded as onco-cytic. If the lesion did not qualify for one of these“metaplastic” lineages,23 or if it resembled the GB epi-thelium, it was then classified as biliary (Fig. 7).

The existence of any cell lineage morphology wasacknowledged regardless of the amount, even if it wasminimal or underdeveloped. However, each case was as-signed a final lineage type on the basis of the predominantpattern (> 75% of the lesion).

Immunohistochemical Analysis

Cell Lineage Markers

Immunohistochemical analysis was performed withcell lineage markers, which are known to be differentiallyexpressed in different components of the gastrointestinal(GI) tract and which have also been used for sub-classification of pancreatic IPMNs, ampullary IAPNs, andbiliary IPNs: MUC1, marker of pancreatobiliary differ-entiation10,23,25,40,125–128; MUC2, intestinal (goblet cell)differentiation23,125,127–132; CDX2, intestinal transcriptionfactor23,130,133–135; MUC5AC, foveolar mucin marker (alsopositive in most IPMNs)39,65,130,136,137; MUC6, pyloricmarker, also positive in pancreatic ITPNs124,130,138; Hep-Par, detected in oncocytic IPMNs139–141; CK747,142–146; andCK20.47,142–144,146,147

MethodologyImmunohistochemical analysis was performed using

a polymer-based detection system (Envision+; Dako,Carpinteria, CA) with mouse monoclonal antibodies ac-cording to the manufacturer’s instructions. Sections weredeparaffinized and rehydrated with deionized water.Then, they were heated in citrate buffer, pH 6.0, using anelectric pressure cooker for 3 minutes at 12 to 15 poundsper square inch at approximately 1201C and cooled for10 minutes before immunostaining. All slides were loadedonto an automated system (Autostainer; Dako) in whichthey are exposed to 3% hydrogen peroxide for 5 minutes,

incubated with primary antibody for 30 minutes, in-cubated with labeled polymer (Envision+ dual link) for30 minutes, incubated in 3,30-diaminobenzidine as achromogen for 5 minutes, and counterstained with hem-atoxylin for 5 minutes. These incubations were performedat room temperature. Between incubations sections were

FIGURE 3. A spectrum of papillary and tubular patterns can be seen in ICPNs, often in a mixture (A–D).

FIGURE 4. Intestinal phenotype in ICPN. Morphologically thelesion is highly similar to colonic adenomas, showing pseudo-stratified cigar-shaped nuclei and overall basophilia. This casewas also positive for CDX2 and MUC2.

Am J Surg Pathol Volume 36, Number 9, September 2012 Intracholecystic Papillary-Tubular Neoplasms (ICPN)

r 2012 Lippincott Williams & Wilkins www.ajsp.com | 1283

8/16/2019 Adsay icpn 2012

6/23

washed with Tris-buffered saline. Coverslipping was per-formed using the Tissue-Tek SCA coverslipper (SakuraFinetek USA Inc., Torrance, CA). Positive controls andnegative controls with primary antibody replaced by Tris-buffered saline were run with the patient/study slides. Thedetailed specifications of the antibodies are providedin Table 2.

Evaluation of Immunohistochemical StainsThe percentage of cells showing cytoplasmic(MUC2, MUC5AC, MUC6, CK7, CK20), apical mem-branous or cytoplasmic (MUC1), and nuclear (CDX2)labeling were evaluated by 4 pathologists for extent;

labeling below 10% was designated focal, 10% to 50%moderate, and >50% diffuse.

InvasionThe invasive carcinoma components were classified

according to the WHO and were staged according to theAJCC, 2010. The size of invasive carcinoma (i-size) wasrecorded separately, and invasive carcinomas were alsoclassified as focal if they were r5 mm in size, substantialif 6 to 29 mm, and extensive if Z30 mm.

Statistical AnalysisPatient demographic and clinical characteristics

across the 3 study groups (noninvasive ICPN, ICPNwith invasion, and pancreatobiliary-type invasive ad-enocarcinoma without ICPN) were compared using w2

tests. Overall 1-, 3-, 5-, and 10-year survival in the 3groups was examined using the life tables method. Afterlife table analyses, we constructed Kaplan-Meier survivalcurves accompanied by the corresponding overall andpairwise log rank tests for statistical significance. A sep-arate set of analyses evaluated the distribution of histo-logic ICPN characteristics (growth pattern, cell lineage,and extent of HGD) and compared invasive and non-invasive lesions using w2 tests. The level of significance forall statistical tests was set using a cutoff of

8/16/2019 Adsay icpn 2012

7/23

is, 6.4% of GB carcinomas arose in association withICPN. Of the remainder, 57% had associated overtnontumoral intraepithelial neoplasia.

The female predominance characteristic of GBpathologies was not as striking in ICPNs: F/M = 2.1, asopposed to 3.9 for pancreatobiliary-type GB ad-enocarcinomas (Table 3). The mean age of the patientswas 61 y (range, 20 to 94 y) as compared to 64 y in pan-creatobiliary-type GB adenocarcinomas.

Among the 62 patients with adequate clinical in-formation accessible to the authors, almost half presentedwith right upper outer quadrant pain, and in the otherhalf the lesion was detected incidentally. Twenty-twopercent of the patients had other neoplasms, most beingGI tract malignancies (gastroesophageal carcinomas, 2;colon carcinoma, 3; pancreatic carcinoma, 2; pancreaticmucinous cystic neoplasm with moderate dysplasia, 1;hepatocellular carcinoma, 1; endometrial carcinoma, 2;uterine cervical carcinoma, 1; prostate carcinoma, 1).

In preoperative radiologic evaluation, almost half of the cases were thought to have GB cancer, whereas about

a quarter were documented to have a polypoid tumor,and in about 10%, the lesion was missed.

In the cases from Chile, the exact frequency of gallstones could not be determined, because in many casesthe stones had been removed and given to the patient’sfamily before the submission of the specimen to thepathology laboratory, and this occurrence was not reli-ably documented in the patient charts. In the cases fromKorea and the United States, only 20% of the properlydocumented cases had gallstones (12/58).

Macroscopic Findings

ICPNs were characterized by prominent exophyticgrowth within the GB (Figs. 2A, B) or by granular, friablesoft-tan excrescences (Fig. 2C). Some cases, in particularthe more granular (papillary) ones, were sessile and morebroad based (Fig. 2D), whereas others, especially thelobulated ones, were pedunculated and had such thinstalks that the lesions often readily detached from thesurface. In fact, for this reason, in many cases, the lesionwas mistaken as sludge or debris in the lumen and notadequately sampled during initial macroscopic examina-tion; their nature was recognized only after the secondround of sampling. Larger lesions often had a hemor-rhagic or necrotic appearance.

The average tumor size was 2.6 cm, and median was2.2 cm (range, 1.0 to 7.7 cm). Seventy percent of the caseswere recorded to be solitary and 30% as multifocal. Skipareas of normal mucosa could be identified within someseemingly solitary lesions. The most common locationswere fundus and body (88% of the cases).

Those cases associated with invasive carcinomahad thickened fibrotic walls; however, as expected,this was difficult to distinguish from ordinary chroniccholecystitis.

Microscopic Findings

General CharacteristicsICPNs were characterized by intraluminal growth

of back-to-back epithelial units, either in a papillary or ina tubular configuration, or both, with minimal or no in-tervening stroma (Fig. 3). The base of the lesions, innoninvasive components, was often sharply demarcatedbecause of the intramucosal nature of the process; how-

ever, in some cases there was extension of the lesion intothe Aschoff-Rokitansky sinuses, creating invaginationsand a pseudoinvasive appearance. Fifty-one percent of the cases had flat-type dysplasia in the adjacent mucosathat merged with the exophytic process.

Transition from low-grade dysplasia to HGD (ad-enoma-carcinoma sequence) in the lesion was evident inmost cases (Figs. 8A, B). Invasive carcinoma (see below),if present, was mostly at the base of the lesion (49 cases;72%) (Fig. 9); however, in 6 cases (9%) it was localized inthe head of the polyp, whereas 8 cases (12%) had invasionin both the base and the head of the lesion. In addition, 5cases (4%) had invasion elsewhere, separate from the

ICPN site.In 31 cases (25%), particularly in those with a

papillary configuration, biliary lineage (see below), andHGD, polymorphonuclear leukocytes showed prefer-ential distribution in and around the epithelium, sparingthe stroma. Twenty-seven cases (22%) also had lympho-plasmacytic infiltrates. Fibrosis and chronic inflammationwere common, but whether this was due to the lesionitself or other instigators such as gallstones could not bedetermined. Interestingly, 13 cases had prominent fol-licular cholecystitis in the adjacent GB. In pyloric com-plex nonmucinous type, the uninvolved GB was oftendevoid of chronic changes.

TABLE 2. Specification of the Antibodies

Antibody Clone Dilution Antibody Source Retrieval Method

MUC1 Ma695 1:160 Leica Microsystems, Bannockburn, IL CitrateMUC2 CCp58 1:100 Leica Microsystems CitrateMUC5AC CLH2 1:200 Leica Microsystems TrilogMUC6 CLH5 1:80 Leica Microsystems Trilog

CK7 OB-TLI2/30 1:40 Dako CitrateCK20 Ks20.8 1:40 Dako CitrateCDX2 CDX2-88 1:200 Biogenex, San Ramon, CA CitrateHEPPAR OCHIE5 1:160 Dako Citrate



8/16/2019 Adsay icpn 2012

8/23

ArchitectureAlthough 90% of the ICPNs showed a mixture of

papillary and tubular areas, 53 cases (43%) qualified aspapillary, 32 cases (26%) as tubular, and 37 cases (31%)as tubulopapillary when cutoffs of 75% and 25% wereused. The mean size of the papillary group was 2.8 cm,tubulopapillary 2.7 cm, and tubular 2.0 cm.

Papillary cases tended to be of either biliary (64%)(Fig. 7) or mixed type showing some biliary features(15%) and rarely of the gastric or intestinal type. Incontrast, tubular cases were more of the gastric lineage(84%) (Fig. 5) or mixed with gastric-type features. Ninetypercent of the tubulopapillary cases had mixed cellularlineage.

Extensive HGD appeared to be more common inthe tubulopapillary (68%) and papillary groups (55%)than in the tubular group (32%) (P = 0.008). Comparedwith neoplasms without invasion those with invasionincluded higher proportions of papillary (47% vs. 38%)and tubulopapillary (38% vs. 22%) lesions and a lower

number of tumors with tubular growth pattern (15% vs.40%). As shown in Table 4, these differences were stat-istically significant (P = 0.005). In contrast, even casesthat may have qualified as “papillary adenocarcinoma”(papillary pattern with extensive HGD) exhibited low-grade dysplastic foci in 17%, and, furthermore, 32% of these cases were noninvasive.

DysplasiaThe ICPNs exhibited a spectrum of neoplastic

transformation ranging from epithelium that was vir-tually indistinguishable from normal glands to those with

severe cytologic atypia and all the attributes of carci-nomatous transformation (HGD/CIS). This spectrumalso often occurred within a given case (Fig. 8). Thedysplastic nature of the lesion was readily identified bycytologic atypia in many cases. In others, in particularthose with the pyloric mucinous phenotype (see below),the neoplastic/dysplastic nature of the process was de-

termined by the sheer size and the compact back-to-backgrowth of glands with no intervening stroma, akin topyloric gland adenomas or serrated adenomas of theGI tract, which are defined as preinvasive neoplasmsalthough they lack the conventional cytologic atypiaof dysplasia.

HGD was manifested in different patterns. In manycases, pseudostratification of cells with nuclear pleo-morphism and dyspolarity was the main evidence (Figs. 8,9). In others, the epithelium formed a single layer, butshowed substantial nuclear anomalies and clear cell fea-tures with centrally located nuclei. In a smaller subset,HGD was characterized by the cribriform arrangement of

the cells, some with clear cell features. A few examplesshowed solid areas and even exhibited comedo-like ne-crosis (akin to those seen in ITPNs of the pancreas) fo-cally raising the question of surface cancerization by anunderlying invasive carcinoma. However, these foci wereeither far away from the invasive component of the lesionor localized in an otherwise typical preinvasive lesion, orthe case did not have any invasive component.

HGD was more extensive in the cases with asso-ciated invasive carcinoma compared with noninvasiveones (P < 0.001). Using the definitions set forth in theMaterials and Methods section, the proportions of focal,substantial, and extensive HGD among noninvasive

TABLE 3. Comparison of Demographic and Main Clinical Characteristics Among ICPN and Pancreatobiliary (PB)-type GBCarcinoma Cases

ICPN (N = 123)

ICPN Without Invasive

Carcinoma (N = 55)

ICPN With Invasive

Carcinoma (N = 68)

PB-type

Carcinoma (N = 411)

N % N % N % P wAge (y) 0.123

< 55 19 35 16 24 86 2155-64 11 20 22 32 101 2565-74 15 27 13 19 130 32Z75 9 16 12 18 87 21Unknown 1 2 5 7 7 2

Sex 0.037Male 17 31 21 31 84 20Female 36 65 44 65 319 78Unknown 2 4 3 4 8 2

T stage* < 0.001T1 NA NA 22 32 35 9T2 NA NA 32 47 141 34T3 NA NA 14 21 222 54Missing NA NA 0 0 13 3

Overall size 0.151< 3 cm 39 71 38 56 237 58Z3 cm 16 29 30 44 174 42

Invasion size < 0.001< 2 cm NA NA 46 68 140 34

Z2 cm NA NA 21 31 271 66Unknown NA NA 1 1 0 0

*Applies to invasive disease only.wOn the basis of w

2test, calculations exclude subjects with unknown information.

NA indicates not applicable.



8/16/2019 Adsay icpn 2012

9/23

tumors were 24%, 45%, and 31%, respectively. In con-trast, the corresponding percentages for cases with asso-ciated invasive carcinoma were 6%, 23%, and 71%(Table 4).

Cell LineageUnlike IPMNs of pancreas or IAPNs of ampulla,

the ICPNs appeared to have more cell lineage diversity(Fig. 10), rendering the cell lineage–based classificationmore difficult to apply. Transitional forms, mixed areas,and unclassifiable patterns were identified to some degreein 90% of the cases: 76% had at least some foci with

biliary-like features, 72% exhibited at least some areasrecognizable as gastric differentiation, and patterns re-sembling but not entirely qualifying for intestinal differ-entiation were noted in 42%.

When the cases were classified on the basis of thepredominant (> 75%) pattern or overall pattern, thefollowing categories were discerned:

Biliary (n = 61; 50%): The vast majority (95%)had a papillary growth pattern (> 25% of the lesion);56% were in the papillary category, and 39% in the tu-bulopapillary group.

These commonly showed carcinomatous trans-formation: 67% had extensive HGD, and 69% had asso-

ciated invasive carcinoma. Invasive carcinoma in this groupwas focal in 14, substantial in 24, and extensive in 4 cases.

Typically, the cases in this group, by definition, hadmore cuboidal cells (Figs. 7, 11A). Some had clear cellfeatures, and some had solid or cribriform patterns as well.Some had more columnar cells resembling the intestinalphenotype but with more acidophilic cytoplasm and lack of MUC2/CDX2 expression. Commonly, a foveolar compo-nent (with MUC5AC expression) was encountered withinthe lesion or in the background mucosa. Tumor-infiltratinginflammatory cells appeared to be more common in thisgroup: the frequency of prominent neutrophilia was 36%(vs. 14% in others) and that for lymphoplasmacytic in-

filtrates was 28% (vs. 15% in others).A small subset in this category was characterized by

delicate papillary lesions lined by innocuous cells similarto normal GB epithelial cells. These cases, which pre-sumably correspond to biliary-type (adenomas) papil-lomas reported by Albores-Saavedra33 had no HGD orany invasive carcinoma. Select photos of these cases wereshared with Dr Albores-Saavedra, and it was verified thatthese cases match those described by him.

In 8 cases, there were focal features characteristic of cholesterol-type polyp associated with the lesion (in 6,without any cholesterolosis in the uninvolved mucosa),raising the possibility that some of these lesions may be

FIGURE 8. Spectrum of dysplasia (cytoarchitectural atypia) in ICPNs. Low-power examination (A) shows the architecturalcomplexity with disorganized and gland-in-gland appearance. B, Low-grade dysplastic cells with virtually no cytologic atypia isseen adjacent to high-grade dysplastic cells, which reveal stratification, loss of polarity, mucin depletion, disorganization, nuclear enlargement, and moderate nucleomegaly with chromatin clumping.



8/16/2019 Adsay icpn 2012

10/23

arising within cholesterol polyps, as observed in a fewcases in the literature.148–150

The biliary group typically showed CK7 (100%)and MUC1 expression (74%) (Fig. 11B), particularly inthe areas with HGD. Most of them also showed weakexpression of other lineage markers (focal MUC2 in 15%,

focal CDX2 in 15%, CK20 in 28%, MUC5AC in 50%,and MUC6 in 32%).

Gastric (n = 44; 36%): Gastric lineage was oftendetected as a recognizable component in the biliary-typecases but not as much in the intestinal type, akin to whatis observed in pancreatic IPMNs.2,23 However, as in the

pancreas, this may be partly definitional because oncegastric-type cells become HGD and lose their mucin, theyacquire biliary-type morphology (Fig. 8) or become on-cocytic if they acquire voluminous cytoplasm withabundant mitochondria.

Gastric-type metaplastic changes were also fre-quently observed in the uninvolved epithelium of ICPNcases, which is not surprising as they are very common inthe general population as well.151

Cases classified as the gastric type on the basis of thepredominant or overall pattern were mostly tubular (61%tubular and 25% tubulopapillary).

This was also the group with the least number of

carcinomatous transformations: 29 cases (66%) werenoninvasive, and only 27% had extensive HGD. How-ever, as mentioned above, this may be definitional.

Three interrelated but also distinct subsets wereidentified in this category:

(a) Foveolar (n = 20; 16% of ICPNs): This group ischaracterized by relatively larger elongated glands, oftenwith tubulopapillary architecture, lined by foveolar-likecells with abundant pale cytoplasm (Figs. 5, 11C) andperipherally located nuclei. This is the group that ap-peared to be closely related to the biliary type and wasoften associated with HGD changes (95%) and invasivecarcinoma (55%).

MUC5AC was positive in all cases (100%)

(Fig. 11D), and on occasion MUC6 could also be positive(54%). MUC1 was less common and detected mostly inthe areas with HGD.

FIGURE 9. Invasive carcinoma arising in ICPNs. Althoughmost invasive carcinomas are ordinary (pancreatobiliary-type)adenocarcinomas, some are unusual types, such as colloid (asshown here) or neuroendocrine.

TABLE 4. Comparison of Histologic Characteristics Among ICPN Cases With and Without Invasive Carcinoma

Disease Group

ICPN Without Invasive Carcinoma (N = 55) ICPN With Invasive Carcinoma (N = 68)Patient and Disease

Characteristics N % N % P z

Histologic growth pattern 0.005Papillary 21 38 32 47Tubular 22 40 10 15Tubulopapillary 12 22 26 38

Cell lineage < 0.001Biliary (including oncocytic)* 22 40 47 69Gastric foveolar 9 16 11 16Gastric pyloric 20 36 4 6Intestinal 4 7 6 9

Extent of HGD (%)w < 0.001Focal (< 25) 12 24 4 6Substantial (25-75) 23 45 15 23Extensive (> 75) 16 31 47 71

*Oncocytic cases grouped together with biliary ones, as they showed similarity in aspect of HGD and MUC1 expression.wExcludes 6 persons with no HGD.zOn the basis of w2 test.



8/16/2019 Adsay icpn 2012

11/23

(b) Pyloric, simple, mucinous (n = 13; 11% of ICPNs): This group (Figs. 6, 11E) was characterized bytightly packed, evenly sized, small, and bland-appearingpyloric-type (Fig. 6; 8 cases) or Brunner-like (Fig. 11E5 cases) glands showing abundant apical mucinous cyto-plasm, peripherally located nuclei, low nucleo-cytoplas-mic ratio, and minimal or no intervening stroma. One of

the cases in this category was associated with a Brunnergland hamartoma in the duodenum, suggesting a genetictendency to form abnormal Brunner gland–type pro-liferations. This group had extensive HGD in only 1 caseand invasive carcinoma in 2. It is noteworthy that in 1case, invasive carcinoma was distant from the polyp.MUC6 expression (Fig. 11F) was strong in 11 (90%) andfocal in 2 cases.

(c) Pyloric, complex, nonmucinous (n = 11; 9% of ICPNs): This was a highly distinctive group characterizedby complex growth of small, nonmucinous tubular units(Figs. 11G, 12) that diffusely and uniformly expressedMUC6 (100%) (Fig. 11H). The tumors typically formed

pedunculated multinodular polyps that were prone todetachment from the surface and often developed hem-orrhagic material in the lumen mistaken as “sludge.”These were typically composed of more complex tubularunits, which often showed irregular and mildly variegatedglandular architecture associated with cystic dilatation of the glandular elements, and some cysts containing gran-

ular acidophilic secretory material (Fig. 12). Typically,the lobules were covered by normal-appearing epithelium.The nucleo-cytoplasmic ratio in this complex group wasfairly high, the nuclei were cuboidal, and nucleoli wereoften visible (Fig. 13). Many cases (5/11; 45.5% of thegroup) showed foci with overlapping optically clear nuclei(Fig. 14). Paneth-type cells and cells with endocrine typegranules were also seen in certain cases with this pattern,and were prominent in some areas. Amyloid-like hyali-nization of the stroma was noted in some cases. Squa-moid morule formation (Fig. 15), represented as solidsquamoid clusters of meningothelial-like cells in awhorled configuration, some with optically clear nuclei

FIGURE 10. Mixed cell lineages and hybrid or difficult-to-classify patterns are common in ICPNs. In many cases of the biliary type,the cells focally exhibit columnar morphology and mimic intestinal differentiation (A and C). In some others (B), the cells arecuboidal and show biliary-type pattern, but the mucin content and presence of goblet-like cells make it difficult to place this lesioninto a specific type. In some of the tubular examples, the surface component shows a different cytology with stratification of cells(D).



8/16/2019 Adsay icpn 2012

12/23

resembling inclusions, was noted in a significant pro-portion (64%) of these cases. This was not seen in any of the other categories.

It is noteworthy that cases in this group appeared to bestrikingly uniform, and although they were often large tu-mors, they rarely showed mucinous-type gastric lineage in

the background, nor transition to any of the other lineages.In addition, the uninvolved GBs were mostly devoid of anypathology, including the conspicuous lack of gastric meta-plasia. These cases, particularly the ones with excess mucindepletion, resemble, in terms of morphology and uniformity,ITPNs of the pancreas.2,28,38–41 Immunohistochemically,diffuse and strong MUC6 expression was uniform in thisgroup (100%) (Fig. 11H). Focal MUC1 expression wasidentified in 4 (57%) cases, mostly limited to the areas of HGD. MUC5AC was expressed in 42%, CK7 in 100%,and others were mostly negative.

The frequency of associated invasive carcinoma wasquite low in this group (18%).

FIGURE 12. Pyloric complex nonmucinous type is a distinctsubset that tends to have large pedunculated polyps. Thepolyps are composed of a nodular pattern of tightly packedsmall to medium sized tubules, some showing cystic dilatationcontaining granular secretory material.

FIGURE 11. Spectrum of cell lineages in ICPNs. Biliary type is characterized by cuboidal nondescript cells (A) and common (74%)MUC1 expression (B). Foveolar variant typically shows large elongated glands lined by tall columnar mucinous epithelium withpale nonchromophilic cytoplasm (C) and consistent MUC5AC expression (D). Pyloric simple mucinous type reveals back-to-backsmall tubular units that resemble both pyloric gland adenomas and Brunner gland adenomas (E) and reveals consistent MUC6 (F)positivity. Pyloric complex nonmucinous type are defined by a distinctive lobulated growth pattern of small tubular unitsillustrated in Figures 13 and 14, which, at the cytologic level, correspond to relatively uniform cuboidal cells with nonmucinouscytology (G), some showing nuclear features of papillary thyroid carcinoma including clearing and overlapping. MUC6 is con-sistently expressed in this group (H) in virtually all cases. Intestinal lineage in ICPNs (I) is similar to colonic adenomas or intestinal-type IPMNs and exhibit common (75%), albeit not uniform, expression of CDX2 (J). Oncocytic examples of ICPN (K) areidentified by arborizing papillae that are lined by oncocytic cells, although immunophenotypically they are different from their pancreatic/biliary counterparts by lack of HepPar, which was detected in only 1 case (L).



8/16/2019 Adsay icpn 2012

13/23

Intestinal (n = 10; 8% of ICPNs): These cases weredistinguished from other types by overall basophilia evi-dent even at low magnification, created by nuclearcrowding of the pseudostratified cells with enlarged,cigar-shaped nuclei and dense cytoplasmic chromophilia(Figs. 4, 11I). In many cases, the lesions were easily dis-

tinguishable from the adjacent uninvolved mucosa bytheir clearly dysplastic nature. Immunohistochemically,these were commonly positive for either CK20 (100%) orCDX2 (75%) (Fig. 11J). MUC2 positivity was seen in50%. The other markers were typically negative exceptCK7 (Table 5).

Invasive carcinoma was seen in 6 cases. Four wereof pancreatobiliary type, and 2 were colloid type. Inter-

estingly, death from disease was not observed in any of the intestinal type cases, but the number of cases was toosmall to derive a conclusion.

Oncocytic (n = 8; 6% of ICPNs): These cases weredistinguished by the complexity of the papillae and on-cocytic cytology (Figs. 11K, 16).30,34,48,130,139–141,152 Theydiffered, however, from their pancreatobiliary counter-parts by the high degree of atypia, dirty background, andlack of HepPar labeling (only 1/6) (Fig. 11L). Moreover,only 1/6 showed MUC6 expression (Table 5). Conversely,they all expressed MUC1, which is variable in oncocyticIPMNs. Because of the immunophenotypic similarities tobiliary type, for the purposes of comparative analysis

(Table 4), we grouped these oncocytic cases along withbiliary type. Two had eosinophilic intracytoplasmicglobules very similar to the “Mallory” globules seen inrenal cell carcinomas (Fig. 16).153 Invasive carcinoma wasseen in 5 cases.

Clinicopathologic Comparison of Cell Lineage TypesThere were statistically significant differences in the

frequency of invasion among the different cell lineagegroups, with the biliary type showing a significantlyhigher association with invasive carcinoma comparedwith the gastric type (P < 0.001; Table 4).

FIGURE 13. Pyloric complex nonmucinous type. The cells arecuboidal, often with round nuclei and visible nucleoli andmoderate to low amount of nonmucinous (or not overtly

mucinous) cytoplasm. The cytologic features often qualify theprocess as HGD as seen here. See Figure 14 for another cy-tologic pattern seen in this type.

FIGURE 14. Pyloric complex nonmucinous type is charac-terized by a distinctive low-power growth pattern illustratedin Figures 12 and 13, which corresponds to back-to-back smalltubular units that are well formed with open lumina and rel-atively monotonous cuboidal lining, some with prominentnucleoli. Others, such as this case, are more complexand show nuclear features reminiscent of papillary thyroidcarcinomas with overlapping, elongation, and a chromatinclearing.

FIGURE 15. Morules (squamoid corpuscles), some with opti-cally clear nuclei, are encountered commonly in the gastricpyloric types of ICPNs, but not seen in other types. On closeinspection, some of these morular cells show meningothelialfeatures and optically clear nuclei.



8/16/2019 Adsay icpn 2012

14/23

ImmunohistochemistryThe group of cases with prominent hybrid and

difficult-to-classify morphology (56%) also showed acomplex immunohistochemistry (IHC) profile: CK7,100%; MUC1, 64%; MUC5AC, 55%; MUC6, 49%;MUC2, 12%; CDX2, 14%; and CK20, 31%. In most

cases, however, the cell lineage morphology in differentcomponents of the same lesion had relatively predictablecorrelation with the immunophenotype. This was alsomanifested in the correlation of the “final” diagnosis (onthe basis of the predominant or favored pattern) for eachcase as discussed above (also see Table 5). IHC was alsohelpful in distinguishing mimickers, such as the intestinal-like appearance in biliary cases as illustrated in Figures10A, C, from true intestinal differentiation.

Invasive Carcinomas Arising in ICPNInvasive carcinoma was seen in 68/123 (55%) of the

ICPN cases. Conversely, an ICPN component was iden-

tified in 39 of 606 systematically analyzed invasive GBcarcinomas (6.4%). Most of these (87%) were pan-creatobiliary-type adenocarcinomas of the GB. However,9 cases showed other types of carcinoma (mucinous, 4;adenosquamous, 1; neuroendocrine, 2; signet ring, 1;medullary, 1) (Fig. 9, mucinous). It is noteworthy that apatient that had been originally classified as noninvasivewas found to have minute invasion in the recuts obtainedfor this study but not in the original slides.

When noninvasive and invasive cases were com-pared, no age or gender difference was found in these 2groups (Table 3). Cases with associated invasive carci-noma were characterized by predominance of papillary ortubulopapillary growth patterns (85%), biliary cell line-age (69%), and extensive HGD (71%) in their non-invasive component. Although the mean size of thepreinvasive lesion was slightly larger in the invasive group(2.8 vs. 2.4), overall size of the tumors were not sig-nificantly different between these 2 groups (P = 0.22).

Invasive carcinoma was focal (r5 mm) in 20/68,

substantial (6 to 29 mm) in 39/68, and extensive (Z30mm)in 9/68 cases. Twenty-two cases were staged as T1 byAJCC, 32 were T2, and 14 were T3. Invasive carcinomasarising in ICPNs were significantly lower-stage tumors atthe time of diagnosis compared with pancreatobiliary-type GB carcinomas (Table 3). This, however, did notseem to be the reason for their favorable outcome (seenext section).

PrognosisThe overall survival after diagnosis of ICPN was

very good. Patients with noninvasive ICPNs had 1-, 3-,and 5-year survival rates of 90%, 90%, and 78%, versus

69%, 60%, and 60%, respectively, of those with asso-ciated invasive disease. Even cases with associated in-vasive carcinoma appeared to have a significantly betterprognosis than pancreatobiliary-type GB carcinomaswith median survivals of 35 versus 9 months. As shownin Figure 17, the differences in survival across the 3groups were statistically significant in both overall andpairwise analyses (all P < 0.001). To further investigatewhether this survival difference was dependent onthe stage, we performed a stage-matched analysis, and thesurvival difference between 2 groups was still found to bestatistically significant (P < 0.0001) (Fig. 18). To confirmthe stage independence of this survival difference, Cox

TABLE 5. Immunohistochemical Analysis of ICPN Types (Data are Shown as %)

All ICPNs Biliary Gastric Pyloric Gastric Foveolar Intestinal Oncocytic

MUC1 64 74 35 69 25 100MUC2 12 15 6 8 50 0MUC5AC 55 47 44 100 0 50MUC6 49 32 94 54 25 17

CK7 100 100 100 100 100 100CK20 31 28 19 25 100 50CDX2 14 16 6 8 75 0HepPar NA NA NA NA NA 17

NA indicates not applicable.

FIGURE 16. Oncocytic-type ICPNs are characterized bycomplex/arborizing papillae lined by 2 to 5 cell layers of cu-boidal cells showing abundant acidophilic granular cytoplasmand single prominent nucleoli. Intraepithelial lumen formationis also noted. Mallory-like hyaline bodies documented in renalcell carcinomas can be encountered. It should be noted herethat immunohistochemically these oncocytic ICPNs do notshow the characteristic phenotype of their pancreatobiliarycounterparts (they lack HepPar and MUC6 and show diffuseMUC1), which places them closer to biliary-type ICPNs.



8/16/2019 Adsay icpn 2012

15/23

proportional regression and hazard ratio analysis wereperformed comparing the 2 groups with and withoutstage matching. The hazard ratio was statistically sig-nificant in both cases [0.372 (95% confidence interval,0.238-0.581; P < 0.0001) without the matching vs. 0.538(95% confidence interval, 0.342-0.845; P = 0.007) withthe matching] thus confirming that the survival differencewas independent of the stage.

Among 55 patients with noninvasive ICPNs, 4 diedof cancer, whereas 1 died at 14 months; the remaining 3

died >5 years after their diagnosis (63, 84 and 94 mo,respectively; median 72 mo). Three of these cases werediagnosed with biliary tract cancers on the basis of imaging studies and signs of biliary obstruction. Thefourth one was recorded to have died of “stomach can-cer” although no biopsy was performed. These cases had

been sampled in 21, 12, 30, and 8 blocks, respectively. Allbut the last one was submitted entirely. See Table 6 forother characteristics of these cases.

DISCUSSION

Definition of ICPNThese tumoral intraepithelial neoplasms of the GB

are remarkably analogous to IPMNs and ITPNs of thepancreas,27,38,41 as well as biliary IPNs 20,22,40,42,48,49,139,154

and IAPNs,54 in their exophytic nature, expression of cel-lular lineages (biliary, gastric, intestinal, oncocytic), andthe presence of a spectrum of dysplastic change (adenoma-carcinoma sequence), all often occurring in varying de-grees, thus warranting their evaluation in a parallelcategory. For this category, we hereby propose the termICPNs to embrace all mass-forming, preinvasive neoplasiaZ1.0 cm, recognized in the WHO-2010 classification as“adenoma” (tubular, papillary, pyloric gland, foveolar

gland, biliary, intestinal, or otherwise) and “intracysticpapillary neoplasms” (intestinal or pancreatobiliary).1 Asillustrated in this study, the overlap between these subsetsare too great (perhaps with the exception of pyloric type),and, more importantly, a significant proportion of thecases cannot be readily placed into one of the WHO cat-egories (Table 7), further warranting their collection under1 unified entity.

The criterion of Z1 cm was chosen to define thiscategory because this is widely used by surgeons and radi-ologists to determine indication for cholecystectomy, and itwas also used by the WHO-2010 to define tumoral intra-epithelial neoplasm in the pancreatobiliary tract.2,27,40,42,49

In addition, this Z1 cm criterion allows distinction of

mass-forming lesions from their flat counterparts, and,more importantly, it provides a specific guideline, albeitarbitrary, to differentiate these from smaller polypoid py-loric gland metaplasias. It is clear, from both this study andanalysis of the literature, that although the vast majority of these subcentimeter lesions are clinically inconsequential,the larger (Z1 cm) lesions often show cancerous trans-formation.13,57,59,81,92,93,95,97,107,155–157

The term ICPN parallels with their ampulla-pan-creatobiliary counterparts. “Intracholecystic” was chosenby the suggestion of Dr Juan Rosai (personal communi-cation, Washington, DC, March, 2010) in order to specifyboth the localization (intramucosal; in the GB) and the

preinvasive nature of these neoplasms. Their exophyticnature, in contrast, is highlighted by the name “papillary-tubular,” which emphasizes the fact that these tumorsmay have both papillary and tubular configurations oreither one of them, often in a mixture. The name papil-lary-tubular is favored over “tubulopapillary” (or“tubulovillous”), because the latter is a term that definesa specific subset of tumoral intramucosal neoplasms inthe pancreas (ITPNs),2 whereas only a small percentageof ICPNs display features akin to these tumors. Finally,the possibility of incorporating the word “mucinous” inthe category designation was considered but dismissed,because this descriptor was recently dropped from the

FIGURE 17. ICPNs have incomparably better survival thannon–ICPN-associated pancreatobiliary-type adenocarcinomasof the GB. Not surprisingly, ICPNs without invasive carcinomahave a long protracted clinical course with only few deaths

occurring in long-term follow-up. Invasive carcinomas arisingin ICPNs are not as kind.

FIGURE 18. Stage-matched comparison of invasive carcino-mas arising from ICPNs versus ordinary pancreatobiliary-typeadenocarcinomas of the GB that are non-ICPN associated il-lustrates that the survival advantage of the former is largelyretained even within the same stage groups.



8/16/2019 Adsay icpn 2012

16/23

nomenclature of biliary examples of this entity due to thefact that, like in ICPNs, mucin production is only a minorand less common feature in biliary IPNs in contrast withpancreatic IPMNs, which are often characterized by co-pious mucin production.

IncidenceICPNs are rare but their exact incidence is difficult

to determine. There are very limited data in the literature.Moreover, the numbers are conflicting, mostly owing tothe definitional variations, with most studies including

microscopic pyloric gland proliferations (presumablymetaplasias)55,65,68,88,158–160 and others excluding thosethat are invasive. As defined in this study, the incidence of ICPN was found to be 75%) papillary. This findingmay not be surprising, considering that in the GI tractsmaller lesions often prove to be tubular, whereas largerlesions exhibit more papillary (villous) growth, presum-ably because the latter have a tendency to become larger.Also not surprisingly, the frequency and amount of HGDand associated invasive carcinoma are significantly higherin papillary and tubulopapillary cases than in tubularones (see below).

The vast majority of the ICPN cases exhibit amixture of low-grade and high-grade dysplastic foci

within the same lesion. The overall frequency of HGD,however, is very high; only a small minority (5%) of casesis entirely devoid of HGD. Furthermore, HGD is ex-tensive (> 75% of the lesion) in half of the cases at di-agnosis. The frequency of associated invasive carcinomais significantly higher in cases with extensive HGD.

Like their counterparts in the pancreatic and biliarytract, ICPNs exhibit a spectrum of cell lineages that re-capitulate those of the GI tract. Interestingly, unlike theirpancreatic counterparts, in which these lineages are oftenpure and discernible, a mixture of these cell lineages is

seen in coexistence in a significant proportion of ICPNcases. This is also reflected in the heterogenous stainingpattern by IHC.

On the basis of the predominant pattern, the mostcommon cell type is classifiable as biliary and commonlyexpresses MUC1. In addition to resembling GB epi-thelium or pancreatobiliary-type IPMNs, this type alsoexhibits variants including clear cell and more columnarintestinal-appearing cells. The gastric phenotype has 2distinct types: the foveolar type, with uniform MUC5ACexpression, is closely related to the biliary phenotype(often admixed with it) and is also commonly accom-panied by invasive carcinoma (in 60% of the cases). The pyloric type, however, is characterized by diffuse/uniform

MUC6 expression, whether mucinous or nonmucinous,and is often large, homogenous with morule formation,and has a significantly lower frequency of associated in-vasive carcinoma (18%, similar to that of the gastric typeof pancreatic IPMNs). The mucinous ones resemble“pyloric gland adenomas” of the GI tract,37,123,124,162 andthey are also similar to the polypoid pyloric gland meta-plasias in the GB, which appears to be the source of thecontroversies regarding the nature of these lesions in theliterature, because most major studies on “adenomas”were composed predominantly of subcentimeter lesions of this morphology.65,68,73,88,90 The pyloric complex non-mucinous type (Figs. 11G, 13, 14) is quite distinctive in

that it is often associated with a relatively clean back-ground without significant inflammation and formscomplex, pedunculated multinodular intraluminal tumorsthat detach readily, such that they may be dismissed asnecrotic debris in the macroscopy room. Some examplesof this type are similar to pancreatic ITPNs, not onlybecause of tubular growth and MUC6 expression, butbecause of their nonmucinous cytomorphology. In addi-tion, some have nuclear overlapping and chromatinclearing resembling papillary thyroid carcinomas. Moruleformation is most commonly encountered (64%) andseen almost exclusively in this type. Studies have shownthat tumors with morule formation such as endometrial

TABLE 6. Clinicopathologic Features of Patients With ICPNs Without Invasive Carcinoma Who Died of Disease

Sex Age Tumor Size Growth Pattern Cell Lineage Extent of HGD (%) # of Blocks Examined Survival (mo)

F 64 3.1 Tubulopapillary Gastric foveolar 25-75 21 14F 75 2.1 Papillary Gastric foveolar 25-75 12 63F 63 6.2 Tubulopapillary Biliary 25-75 30 84F 82 1.6 Papillary Biliary > 75 8 94



8/16/2019 Adsay icpn 2012

17/23

carcinomas, cribriform-morular variant of papillary thy-roid carcinomas, and fetal-type pulmonary adenocarcino-mas share the common characteristics of occurring morefrequently in women and molecular association with es-trogen-activated b-catenin pathway alterations.163 The

same concept may very well be in play in the pyloric-typeICPNs as well. This issue warrants further investigation.Although few examples of the pyloric complex non-mucinous type appear to be a high-grade transformationof the mucinous type, many are uniformly complex

TABLE 7. Comparison of the Prior Terminology and Concepts in the Literature and the Findings in This Study

Question Previous Literature This Study

What is the behavior of “adenomas”? Impression: all are benign (on the basisof a small number of cases). Average sizein literature B0.7cm

Those that are >1 cm often have HGD,and even the “pyloric gland” type exampleshave 18% invasive carcinoma

What is the behavior of “papillary carcinoma/adenocarcinoma”?

Impression: a variant of adenocarcinoma;all malignant; dismal prognosis similar toordinary GB adenocarcinoma

Many are noninvasive; noninvasive ones,5-yr survival 90%; invasive ones, 60%.

How do we designate a mass-forming preinvasiveneoplasm in the GB?

7 supposedly distinct categories collectedunder 2 generic groups of “adenoma”and “intracystic papillary neoplasm.”Some defined by growth pattern(“tubulopapillary adenoma”), some bycombination of dysplasia+growth(“papillary carcinoma”), some on thebasis of cell lineage (“intestinal adenoma”)

A case can be reported as: ICPN, with(amount of) HGD, (cell lineage) type,(type of) growth pattern

What do I do with this extensive in situ carcinomathat has prominent papillary configuration of tallpapillae, but not forming a distinct mass?

“Papillary adenocarcinoma” CIS, flat type, with prominent papillarypattern (not ICPN)

What do I do if I see a GB case that looks likeintestinal-type IPMN of the pancreas?

Not clear: “adenoma” or “intracysticpapillary neoplasm” or “IPMN of GB”

ICPN, intestinal type

What do I do if I see a tubulopapillary neoplasm thathas intestinal phenotype?

Not clear: “tubulopapillary adenoma” or“intestinal adenoma”

ICPN, intestinal type, with tubulopapillarypattern

What is the clinical outcome of a tubulopapillaryneoplasm that has intestinal appearance?

No literature If noninvasive 5-yr survival 90%; minimalrisk of recurrence at long term

What is the diagnosis and prognosisof a tubulopapillary neoplasm withextensive HGD?

Not clear: if regarded as “tubulopapillaryadenoma,” then benign; if regarded as“papillary carcinoma,” then very aggressive

ICPN with HGD Good prognosis(5-yr survival 90%); however, long-termF/U is warranted

What do I call this papillary neoplasm that hasintestinal features but according to me does not

appear to be intestinal and in which CDX2 andMUC2 are negative?

Not clear ICPN, biliary type

What do I call this large polyp composed of complexbut small glandular units of nonmucinous cells?

Not clear: “pyloric gland adenoma”? ICPN, pyloric complex nonmucinous

What do I call this oncocytic papillary lesion? Howcommon are these?

No literature; not reported IC PN, oncocytic type; 6% of the cases

What do I call a collection of pyloric-like glands thatforms a 4-mm nodule?

“Pyloric gland adenoma” Polypoid metaplasia, not ICPN

What do I call a polypoid lesion that has pyloric-typeglands with intervening fibromuscular stroma?

“Pyloric gland adenoma” Non-neoplastic; not ICPN;fibromyoglandular polyp152

What percentage of invasive carcinomas of the GBarises from mass-forming preinvasive neoplasm?

Wide range; 5%-23%; all small studies 6.4% (on the basis of systematic review of 616carcinoma cases and 3265cholecystectomies)

I have an invasive carcinoma arising from a mass-forming preinvasive neoplasm. Is this going tobehave like an ordinary carcinoma?

Not clear Limited data suggest that they may behavebetter than ordinary invasive carcinomathat arise from flat IN

F/U indicates follow-up; IN, intraepitelial neoplasia.



8/16/2019 Adsay icpn 2012

18/23

without any mucinous component in the lesion or in theuninvolved GB. More importantly, despite their complexarchitecture, large size, and the common presence of HGD, this type of ICPNs is associated with invasivecarcinoma in only 18%.

Convincing intestinal lineage is relatively un-

common in ICPNs (8%). Furthermore, in this study, eventhough stringent morphologic criteria were used, only75% of the cases that were classified as intestinalexpressed CDX2 and 50% MUC2, suggesting that in-testinal differentiation, even when it occurs, is less com-plete than it is in the pancreatic IPMNs23 or IAPNs.54

The fact that the intestinal pathway is significantly morecommon in the pancreatic main-duct IPMNs than inICPNs despite the fact that intestinal metaplasia is morecommon in the non-neoplastic GB (10%)151 may suggestthat intestinal lineage in tumorigenesis may be related tothe reflux of intestinal contents rather than to an em-bryologic remnant phenomenon. Along these lines, the

frequency of intestinal lineage (and other cell types) oc-curring in the ICPNs is fairly similar to those identifiedin the extrahepatic bile ducts in a recent study fromMemorial Sloan-Kettering Cancer Center,139 indicatingthat the biliary tract epithelium (bile ducts and GB alike)is less prone to follow an intestinal pathway than the mainpancreatic duct.

As in other areas of the ampullo-pancreatobiliarytract, the oncocytic phenotype is fairly uncommon inICPNs. More importantly, it does not show the classicimmunophenotypic features of oncocytic IPMNs orIPNs,30,34,48,130,139–141,152 often lacking HepPar andMUC6 expression and consistently showing MUC1 ex-pression instead.

In addition to its aid in verifying the cell lineagesdiscussed above, immunophenotyping also discloses somegeneral characteristics of ICPNs. True to their biliaryorigin, ICPNs are typically positive for CK7, even in thecases that show intestinal differentiation. It is also note-worthy that MUC1, in addition to being a fairly goodindicator of biliary differentiation, is also expressed in thehigh-grade areas of any type and thus may serve as amarker of HGD. It may be important to reiterate herethat a significant proportion of the ICPNs have hybridpatterns and heterogenous areas that coexpress differentimmunohistochemical markers. In fact, this pronouncedproclivity to form hybrid phenotypes further necessitates

the classification of these lesions under 1 umbrellacategory.

Invasive Carcinomas Arising in ICPNsInvasive carcinoma is seen in more than half of the

ICPNs at the time of diagnosis. Most of these are pan-creatobiliary-type GB adenocarcinomas; however, othertypes such as mucinous, neuroendocrine, etc. are also seenand perhaps even with higher frequency than those aris-ing from nontumoral (flat) intraepithelial neoplasia.164

Conversely, when the issue is examined from the invasivecarcinoma perspective, an ICPN component is identifiedin 6.4% of all invasive carcinomas of the GB. In the lit-

erature, a wide range of figures have been quoted for thisassociation, ranging from 5% to 23%,13,15,66,68,73 whichis most likely because of the definitional differencesdiscussed earlier.

As discussed above, factors in the ICPN compo-nent, which appear to have an association with the risk of

invasion include amount of papillary configuration, ex-tent of HGD, and predominance of nonpyloric cell line-ages. Size may also be a factor but did not reach statisticalsignificance in this study. All cases that have an associatedinvasive carcinoma have at least focal HGD. As is thecase in lower GI tract and pancreatobiliary tract, invasivecarcinomas are more likely to develop in the papillarythan in the tubular cases, and this has also been noted inthe literature on studies of “papillary adenocarcinoma” of the GB.57 As in the pancreatic IPMNs and ITPNs,152 thegastric and pyloric lineages in ICPNs appear to have alesser tendency for invasion. However, even the pyloric-type ICPNs with exclusively tubular patterns, which have

long been considered entirely benign innocuous lesions,were found in this study to display HGD frequently, and,more importantly, 18% of these were associated withinvasive carcinoma. This figure, incidentally, is virtuallyidentical to that reported for gastric-type/branch-ductIPMNs.2,152,165–169 In fact, when the literature on the GB“pyloric gland adenoma” cases is analyzed carefully, itbecomes clear that the larger lesions often do show car-cinomatous transformation; unfortunately, this fact hasbeen largely diluted in studies that are composed mostlyof subcentimeter lesions,65,73,88 which may be more ap-propriate to classify as polypoid pyloric gland meta-plasias and common findings in cholecystitis.151

Incidentally, similar pyloric gland lesions in the GI tract

are also found to be associated with invasive carcinomaeven though they are very innocuous appearing.123,124

Biological Behavior and Clinical OutcomeParameters

The prognosis for ICPNs is incomparably betterthan that for other invasive carcinomas of the GB. Notsurprisingly, noninvasive cases have a far better prog-nosis: the 3- and 5-year survival rates are 90% and 78%,versus 60% and 60% in invasive ones (Fig. 17). In con-trast, it would be important to note that some patientswith noninvasive cases also die of cancer. In this study,among 55 noninvasive cases, 4 patients died of cancer, 3

of which were reported as “biliary tract” cancer byimaging and biliary obstruction signs, and these deathsmostly occurred long after the diagnosis of ICPN (median73.5 mo), suggesting that these patients may have sufferedfrom a new primary in the remaining biliary tract. All hadHGD, and 1 had extensive HGD. All had a papillarygrowth pattern. The cell lineage of these cases was gastricfoveolar in 2 and biliary in 2 (Table 6). The fact that somepatients with noninvasive ICPNs die of tumor may not besurprising, as it is now well established that a similarpercentage of noninvasive pancreatic IPMNs also suc-cumb.10,170,171 Furthermore, in nontumoral (flat) CIScases of GB, recurrences and metastasis were seen in a



8/16/2019 Adsay icpn 2012

19/23

similar percentage in spite of being diagnosed by the mostexperienced who ruled out invasive carcinoma by thor-ough sampling.104 In these patients, the tumors weresampled in 8 to 30 blocks, and 3 of 4 were submittedentirely (Table 6). Thus, this adverse outcome of thenoninvasive cases may not necessarily be due to “missed”

invasive carcinomas but rather may be attributable to afield-defect phenomenon, although further studies wouldbe necessary to investigate this possibility. The fact thatmany deaths occurred several years after the original di-agnosis also points toward this explanation. This brings upthe question of surveillance of these patients; however, thebiliary tract still remains difficult to image and highlychallenging to biopsy. Moreover, carcinomas of this regionare often subtle and difficult to distinguish from reactive/regenerative changes. Most importantly, even ICPNs withassociated invasive carcinoma have a better prognosis thanpancreatobiliary-type GB carcinomas (without ICPN), andthis survival advantage was independent of size and

persisted even with stage-matched comparison (Fig. 18).This suggests that ICPN-associated invasive carcinomasmay have distinct biological properties.

Surgical Pathologic Evaluation and ReportingIn evaluation of these lesions, regardless of the

terminology, of foremost importance is to rule out in-vasive carcinoma by thorough sampling and carefulevaluation, considering that the prognosis of noninvasivecases is vastly different from those with associated in-vasive carcinoma. It should be kept in mind that thepolypoid lesions often detach from the mucosa and ap-pear as “sludge” in the lumen of the GB or the container.It should also be remembered that invasive carcinoma

may be grossly inapparent and may also occur away fromthe main ICPN lesion. If identified, invasive carcinomaought to be typed and staged as customary. For thenoninvasive cases, it is important to assess the presenceand amount of HGD, the size of the lesion, the cell lin-eage type, and, when possible, the amount of papillaryconfiguration and the margin status. For cases with hy-brid patterns, immunohistochemical markers MUC1,MUC2, MUC5AC, MUC6, and CDX2 may help tofurther characterize the lineage; whether this is necessaryis debatable, as there are no established management al-gorithms for these lesions yet.

In reporting these lesions, the approach used in

other organs such as the breast is recommended. Ac-cordingly, an example for a report of a noninvasivecase could read as: “Intracholecystic papillary-tubularneoplasm, 4.0 cm, with extensive HGD (> 75% of thetumor), pyloric complex nonmucinous type, with pre-dominantly tubular pattern. No invasive carcinoma isidentified.” For a case with an associated invasive carci-noma, the diagnosis can be rendered as “invasive ad-enocarcinoma (0.5 cm), moderately differentiated, colloidtype, invading into but not through the muscularis(pT1b), arising in an intracholecystic papillary-tubularneoplasm (2.5 cm) of the biliary type with predominantlytubulopapillary pattern.” Characteristics of the invasive

carcinoma such as perineurial invasion are to be docu-mented in further detail.

In conclusion, mass-forming (Z1 cm) preinvasiveneoplasms of the GB share common features and alsoexhibit many similarities to biliary IPNs and pancreaticIPMNs and ITPNs, in that they demonstrate a spectrum

of dysplastic change, variable configuration, and differentcell lineages, often in a mixture. We believe these neo-plasms can be regarded under a generic category forwhich we propose the term ICPN. ICPNs are indolent,and even the invasive cases appear to have a significantlybetter overall prognosis than pancreatobiliary-type ad-enocarcinomas unaccompanied by ICPNs.

ACKNOWLEDGMENTSThe authors would like to thank Dr Juan Rosai for his

recommandation of the term “intracholecystic” for theseneoplasms. The authors also thank Allyne Manzo for herassistance for the photographs and Leslie Ducato and

Rhonda Everett for their assistance in the preparation of this manuscript.

REFERENCES1. Albores-Saavedra J, Adsay NV, Crawford JM, et al. Carcinoma of

the gallbladder and extrahepatic bile ducts. In: Bosman FT,Carneiro F, Hruban R, et al, eds. World Health OrganizationClassification of Tumors Tumors of Digestive System. 4th ed. Lyon:IARC Press; 2010:266–274.

2. Adsay N, Fukushima N, Fukayama M, et al. Intraductal neoplasmsof the pancreas. In: Bosman F, Carneiro F, Hruban R, et al, eds.World Health Organization Classification of Tumors Tumors of

Digestive Systemedn, Vol . Lyon: IARC Press; 2010:304–313.3. Bedenne L, Faivre J, Boutron MC, et al. Adenoma-carcinoma

sequence or “de novo” carcinogenesis? A study of adenomatous

remnants in a population-based series of large bowel cancers.Cancer. 1992;69:883–888.

4. Chen CD, Yen MF, Wang WM, et al. A case-cohort study for thedisease natural history of adenoma-carcinoma and de novocarcinoma and surveillance of colon and rectum after polypectomy:implication for efficacy of colonoscopy. Br J Cancer. 2003;88:1866–1873.

5. Tahara E. Genetic pathways of two types of gastric cancer. IARC Sci Publ . 2004;157:327–349.

6. Fearon ER, Vogelstein B. A genetic model for colorectal tumori-genesis. Cell . 1990;61:759–767.

7. Hermanek PJ Jr., Giedl J. The adenoma-carcinoma sequence inAMMN-induced colonic tumors of the rat. Pathol Res Pract.1984;178:548–554.

8. Sessa F, Solcia E, Capella C, et al. Intraductal papillary-mucinoustumours represent a distinct group of pancreatic neoplasms: aninvestigation of tumour cell differentiation and K-ras, p53 andc-erbB-2 abnormalities in 26 patients. Virchows Arch. 1994;425:357–367.

9. Milchgrub S, Campuzano M, Casillas J, et al. Intraductalcarcinoma of the pancreas. Cancer. 1992;69:651–656.

10. Adsay NV, Conlon KC, Zee SY, et al. Intraductal papillary-mucinous neoplasms of the pancreas: an analysis of in situ andinvasive carcinomas in 28 patients. Cancer. 2002;94:62–77.

11. Aldridge MC, Bismuth H. Gallbladder cancer: the polyp-cancersequence. Br J Surg. 1990;77:363–364.

12. Suzuki A, Takahashi T. Histogenesis of the gallbladder carcinomainduced by methylcholanthrene beeswax pellets in hamsters. Jpn J Surg. 1983;13:55–59.

13. Kozuka S, Tsubone N, Yasui A, et al. Relation of adenoma tocarcinoma in the gallbladder. Cancer. 1982;50:2226–2234.



8/16/2019 Adsay icpn 2012

20/23

14. Goldin RD, Roa JC. Gallbladder cancer: a morphological andmolecular update. Histopathology. 2009;55:218–229.

15. Lee SH, Lee DS, You IY, et al. Histopathologic analysis of adenoma and adenoma-related lesions of the gallbladder. Korean J Gastroenterol . 2010;55:119–126.

16. Castillo-Martin M, Domingo-Domenech J, Karni-Schmidt O, et al.Molecular pathways of urothelial development and bladder

tumorigenesis. Urol Oncol . 2010;28:401–408.17. Hodges KB, Lopez-Beltran A, Davidson DD, et al. Urothelial

dysplasia and other flat lesions of the urinary bladder: clinico-pathologic and molecular features. Hum Pathol . 2010;41:155–162.

18. Cheng L, Davidson DD, Maclennan GT, et al. The origins of urothelial carcinoma. Expert Rev Anticancer Ther. 2010;10:865–880.

19. Pai RK, Mojtahed K. Mutations in the RAS/RAF/MAP kinasepathway commonly occur in gallbladder adenomas but areuncommon in gallbladder adenocarcinomas. Appl Immunohisto-chem Mol Morphol . 2011;19:133–140.

20. Abraham SC, Lee JH, Hruban RH, et al. Molecular andimmunohistochemical analysis of intraductal papillary neoplasmsof the biliary tract. Hum Pathol . 2003;34:902–910.

21. Adsay NV, Adair CF, Heffess CS, et al. Intraductal oncocyticpapillary neoplasms of the pancreas. Am J Surg Pathol . 1996;20:980–994.

22. Abraham SC, Lee JH, Boitnott JK, et al. Microsatellite instabilityin intraductal papillary neoplasms of the biliary tract. Mod Pathol .2002;15:1309–1317.

23. Adsay NV, Merati K, Basturk O, et al. Pathologically andbiologically distinct types of epithelium in intraductal papillarymucinous neoplasms: delineation of an “intestinal” pathway of carcinogenesis in the pancreas. Am J Surg Pathol . 2004;28:839–848.

24. Basturk O, Coban I, Adsay NV. Pancreatic cysts: pathologicclassification, differential diagnosis, and clinical implications. ArchPathol Lab Med . 2009;133:423–438.

25. Furukawa T, Kloppel G, Volkan Adsay N, et al. Classification of types of intraductal papillary-mucinous neoplasm of the pancreas:a consensus study. Virchows Arch. 2005;447:794–799.

26. Hruban RH, Pitman MB, Klimstra DS. Tumors of the Pancreas,Vol. 6. Washington, DC: American Registry of Pathology; 2007.

27. Hruban RH, Takaori K, Klimstra DS, et al. An illustrated

consensus on the classification of pancreatic intraepithelial neo-plasia and intraductal papillary mucinous neoplasms. Am J SurgPathol . 2004;28:977–987.

28. Klimstra DS, Adsay NV, Odze RD, et al. Tumors of the pancreasand ampulla vater. In: Odze R, Goldblum J, eds. Surgical Pathology of the GI tract, Liver, Biliary Tract, and Pancreasedn,Vol . Philadelphia: Elsevier. 2009:909–960.

29. Luttges J, Zamboni G, Longnecker D, et al. The immunohisto-chemical mucin expression pattern distinguishes different types of intraductal papillary mucinous neoplasms of the pancreas anddetermines their relationship to mucinous noncystic carcinoma andductal adenocarcinoma. Am J Surg Pathol . 2001;25:942–948.

30. Rouzbahman M, Serra S, Adsay NV, et al. Oncocytic papillaryneoplasms of the biliary tract: a clinicopathological, mucin coreand Wnt pathway protein analysis of four cases. Pathology.2007;39:413–418.

31. Shibahara H, Tamada S, Goto M, et al. Pathologic features of mucin-producing bile duct tumors: two histopathologic categoriesas counterparts of pancreatic intraductal papillary-mucinous neo-plasms. Am J Surg Pathol . 2004;28:327–338.

32. Tanaka M, Chari S, Adsay V, et al. International consensusguidelines for management of intraductal papillary mucinousneoplasms and mucinous cystic neoplasms of the pancreas.Pancreatology. 2006;6:17–32.

33. Albores-Saavedra J, Henson DE, Klimstra DS. Tumors of theGallbladder, Extrahepatic Bile Ducts, and Ampulla of Vater. Atlas of Tumor Pathology, 3rd Edition, 3 edn, Vol. 27 . Washington D.C.:Armed Forces Institute of Pathology; 2000.

34. Sadakari Y, Ohuchida K, Nakata K, et al. Invasive carcinomaderived from the nonintestinal type intraductal papillary mucinousneoplasm of the pancreas has a poorer prognosis than that derivedfrom the intestinal type. Surgery. 2010;147:812–817.

35. Klo ¨ ppel G, Hruban R, Longnecker DS, et al. Ductal adenocarci-noma of the pancreas. In: Hamilton S, Aaltonen L, eds. World Health Organization Classification of Tumors Tumors of DigestiveSystemedn, Vol . Lyon: IARC Press; 2000:219–230.

36. Amaris J. Intraductal mucinous papillary tumor and pyloric glandadenoma of the pancreas. Gastrointest Endosc. 2002;56:441–444.

37. Kato N, Akiyama S, Motoyama T. Pyloric gland-type tubular

adenoma superimposed on intraductal papillary mucinous tumorof the pancreas. Pyloric gland adenoma of the pancreas. VirchowsArch. 2002;440:205–208.

38. Tajiri T, Tate G, Inagaki T, et al. Intraductal tubular neoplasms of thepancreas: histogenesis and differentiation. Pancreas. 2005;30:15–21.

39. Tajiri T, Tate G, Kunimura T, et al. Histologic and immunohis-tochemical comparison of intraductal tubular carcinoma, intra-ductal papillary-mucinous carcinoma, and ductal adenocarcinomaof the pancreas. Pancreas. 2004;29:116–122.

40. Yamaguchi H, Shimizu M, Ban S, et al. Intraductal tubulopapil-lary neoplasms of the pancreas distinct from pancreatic intra-epithelial neoplasia and intraductal papillary mucinous neoplasms.Am J Surg Pathol . 2009;33:1164–1172.

41. Klimstra D, Adsay NV, Dhall D, et al. Intraductal tubularcarcinoma of the pancreas: clinicopathologic and immunohisto-chemical analysis of 18 cases. Mod Pathol . 2007;20:285A.

42. Zen Y, Fujii T, Itatsu K, et al. Biliary papillary tumors sharepathological features with intraductal papillary mucinous neo-plasm of the pancreas. Hepatology. 2006;44:1333–1343.

43. Oshikiri T, Kashimura N, Katanuma A, et al. Mucin-secreting bileduct adenoma—clinicopathological resemblance to intraductal pap-illary mucinous tumor of the pancreas. Dig Surg. 2002;19:324–327.

44. Lim JH, Jang KT, Rhim H, et al. Biliary cystic intraductalpapillary mucinous tumor and cystadenoma/cystadenocarcinoma:differentiation by CT. Abdom Imaging. 2007;32:644–651.

45. Kim HJ, Kim MH, Lee SK, et al. Mucin-hypersecreting bile ducttumor characterized by a striking homology with an intraductalpapillary mucinous tumor (IPMT) of the pancreas. Endoscopy.2000;32:389–393.

46. Kim H, Lim JH, Jang KT, et al. Morphology of intraductalpapillary neoplasm of the bile ducts: radiologic-pathologiccorrelation. Abdom Imaging. 2011;36:438–446.

47. Kloek JJ, van der Gaag NA, Erdogan D, et al. A comparativestudy of intraductal papillary neoplasia of the biliary tract andpancreas. Hum Pathol . 2011;42:824–832.

48. Ji Y, Fan J, Zhou J, et al. Intraductal papillary neoplasms of bileduct. A distinct entity like its counterpart in pancreas. Histol Histopathol . 2008;23:41–50.

49. Barton JG, Barrett DA, Maricevich MA, et al. Intraductalpapillary mucinous neoplasm of the biliary tract: a real disease?HPB (Oxford). 2009;11:684–691.

50. Kloppel G, Solcia E, Longnecker DS. Histologic Typing of Tumorsof the Exocrine Pancreas, Vol . Geneva: Springer-Verlag; 1996.

51. Sudo Y, Harada K, Tsuneyama K, et al. Oncocytic biliarycystadenocarcinoma is a form of intraductal oncocytic papillaryneoplasm of the liver. Mod Pathol . 2001;14:1304–1309.

52. Zalinski S, Paradis V, Valla D. Intraductal papillary mucinous tumorsof both biliary and pancreatic ducts. J Hepatol . 2007;46:978–979.

53. Zen Y, Sasaki M, Fujii T. Different expression patterns of mucincore proteins and cytokeratins during intrahepatic cholangiocarci-nogenesis from biliary intraepithelial neoplasia and intraductalpapillary neoplasm of the bile duct—an immunohistochemicalstudy of 110 cases of hepatolithiasis. J Hepatol . 2006;44:350–358.

54. Ohike N, Kim GE, Tajiri T, et al. Intra-ampullary papillary-tubular neoplasm (IAPN): characterization of tumoral intraepithe-lial neoplasia occurring within the ampulla: a clinicopathologicanalysis of 82 cases. Am J Surg Pathol . 2010;34:1731–1748.

55. Yamamoto M, Nakajo S, Tahara E. Histological classification of epithelial polypoid lesions of the gallbladder. Acta Pathol Jpn.1988;38:181–192.

56. Adsay NV, Klimstra DS. Benign and malignant tumors of thegallbladder and extrahepatic biliary tract. In: Odze RD, GoldblumJR, eds. Surgical Pathology of the GI tract, Liver, Biliary Tract, and Pancreasedn