- Email: [email protected]
Intestinal and pancreatobiliary differentiation in periampullary carcinoma: the role of immunohistochemistry
Human Pathology (2013) 44, 2213–2219
www.elsevier.com/locate/humpath
Original contribution
Intestinal and pancreatobiliary differentiation in periampullary carcinoma: the role of immunohistochemistry☆ Niraj Kumari MD, DNB, MNAMS a,⁎, Kanchan Prabha MD a , Rajneesh Kumar Singh MS, MCh b , Dinesh Kumar Baitha MSc a , Narendra Krishnani MD a a
Department of Pathology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow 266014, India Department of Surgical Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow 266014, India b
Received 5 February 2013; revised 26 April 2013; accepted 1 May 2013
Keywords: Periampullary carcinoma; Intestinal; Pancreatobiliary; CDX2; Ampullary carcinoma
Summary Periampullary carcinoma (PC) is classified into intestinal and pancreatobiliary subtypes using morphology and immunohistochemistry (IHC). Different combinations of markers have been used in the literature. One hundred eight PCs were classified using morphology and IHC (CDX2, mucin [MUC] 2, cytokeratin [CK] 20, CK7, CK17, and MUC1). The expression of these markers was compared with different histologic subtypes, histopathologic prognostic parameters, and patients' survival. There were 38 intestinal and 53 pancreatobiliary subtypes classified on morphology alone. CDX2 showed high sensitivity (89.5%) and specificity (100%) for intestinal type. CK20 and MUC2 showed low sensitivity (50% and 39.5%) but high specificity (86.8% and 96.2%) for intestinal type. CK7 and CK17 showed a sensitivity of 90.5% and 32% and a specificity of 21% and 89.4%, respectively, for pancreatobiliary subtype. MUC1 was 100% sensitive but 0% specific in pancreatobiliary subtype. The overall median survival in morphologic and IHC intestinal type was 45 months versus 20 months in pancreatobiliary type (P = 0.01). Intestinal and pancreatobiliary types of PC were differentiated in 84.2% of cases by morphology alone and in 87.9% cases with IHC. CDX2-positive tumors had a median survival of 44 months versus 22 months in CDX2-negative tumors (P = .03). IHC helped in reclassifying an additional 4 cases of mixed and other types. Among the panel used, CDX2 showed a high sensitivity and specificity for intestinal subtype and was an independent prognostic marker for longer survival. Thus, CDX2 may be used routinely with morphology in subtyping of PC, and a panel of markers may be used in morphologically difficult cases. © 2013 Elsevier Inc. All rights reserved.
1. Introduction ☆
Funding: The work was funded by the SGPGI Intramural Research Committee of the institute. ⁎ Corresponding author. E-mail addresses: [email protected] (N. Kumari), [email protected] (K. Prabha), [email protected] (R. K. Singh), [email protected] (D. K. Baitha), [email protected] (N. Krishnani). 0046-8177/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.humpath.2013.05.003
The histologic subtyping of periampullary carcinoma into intestinal and pancreatobiliary subtypes is emerging as an important prognostic factor. Despite the common location, 2 different histologic subtypes are associated with different premalignant lesion, cell-type specific markers, oncogene expression, mode of cancer spread, extent of cancers, and
2214
N. Kumari et al.
interaction with extracellular matrix. Westgaard et al [1] suggested that difference in survival of periampullary carcinoma of comparable size is more dependent on histologic differentiation than anatomical origin. The subtyping of periampullary carcinoma into intestinal and pancreatobiliary types was proposed by Albores-Saavedra [2]. He proposed the morphologic criteria that defined intestinal type as cancers composed of well-formed tubular glands, complex cribriform areas, and solid nests, which are indistinguishable from colorectal adenocarcinoma. The tumor cells are pseudostratified tall columnar with oval- or cigar-shaped nuclei located basally and often contain mucin (MUC). The pancreatobiliary type was defined as cancers composed of simple or branching glands and small solid nests of cells surrounded by dense desmoplastic stroma. The tumor cells are cuboidal to low columnar arranged in a single layer without nuclear pseudostratification, and nuclei are rounded with marked variation in size and shape from one cell to the other. Several studies have classified periampullary carcinoma subtypes using the above morphologic criteria alone or in combination of a panel of immunohistochemistry (IHC) markers. The different markers that have been used in the literature to classify these 2 subtypes include cytokeratins (CK7, CK17, and CK20), MUCs (MUC1, MUC2, MUC4, and MUC5AC), and transcription factor CDX2 [3–7]. CK7, CK17, MUC4, and MUC1 have been found to be expressed in pancreatobiliary type of periampullary carcinomas, whereas CK20, MUC2, and CDX2 are expressed in intestinal-type cancers [5]. Paulsen et al [7] used 8 different types of MUCs including MUC1, MUC2, MUC4, MUC5AC, MUC5B, MUC6, MUC7, and MUC8 and found that MUC2, MUC7, and MUC8 were helpful in classifying ampullary carcinoma. The present study aimed to analyze the usefulness of IHC as single or panel of markers that included different CKs, MUCs, and transcription factor CDX2 that would add to the morphologic criteria in subtyping of periampullary carcinoma.
cell carcinoma was used as a positive control for CK17. The antibodies for MUC1 (polyclonal, dilution 1:50) and CK17 (clone E3, dilution 1:40) were obtained from Thermoscientific (San Jose, CA, USA). MUC2 antibody (clone Ccp58, dilution 1:100) was procured from Novocastra (Newcastle, UK), CDX2 antibody (MU392A.UC, dilution 1:50) was procured from Biogenex (Fremond, CA, USA). CK7 (clone OV-TL 12/30, dilution 1:80), and CK20 antibody (clone Ks 20.8, dilution 1:80) was procured from Dako (Glostrup, Denmark). Briefly, the sections were cut at 4-μm thickness and deparaffinized in xylene and graded alcohol. Antigen retrieval was done by microwave treatment using EZ Antigen Retriever (Biogenex) in citrate buffer at pH 6.0 for MUC1, MUC2, and CK17 and in EDTA buffer at pH 8.0 for CDX2, CK7, and CK20 for 30 minutes. Primary antibody was added for a 2-hour duration at room temperature, followed by secondary antibody for 30 minutes (DAKO Real Envision). The slides were then washed and counterstained with Mayer hematoxylin. Positive staining was considered when at least 10% or more tumor cells were stained.
2. Materials and methods
There were 74 men and 34 women (male/female, 2.1:1), with an age range of 27 to 81 years (mean, 57.2 years; median, 58 years). Anatomically, there were 91 ampullary carcinomas, 9 carcinomas of the head of pancreas, 5 distal common bile duct (CBD) carcinomas, and 3 duodenal carcinomas. The histologic subtypes of periampullary carcinoma in different anatomical subsets are given in Table 1. There were 38 (35.1%) cases of intestinal type (Fig. 1A), 53 (49%) cases of pancreatobiliary type (Fig. 1B), 8 (7.4%) cases of mixed type, and 9 (8.3%) cases of other types classified on the basis of morphologic criteria. Mixedtype tumors are classified based on the predominant histologic differentiation, and a defined cutoff has not been proposed; however, in the present study, tumors were classified as mixed based on an arbitrary cutoff of 70:30 ratios, where the intestinal or pancreatobiliary component ranged from 30% to 70%. Any histologic subtype when
One hundred eight consecutive cases of pancreatoduodenectomies were reviewed to classify them as intestinal, pancreatobiliary, mixed, or others subtypes based on the morphologic criteria given by Albores-Saavedra [2]. The different histologic parameters including tumor grade, stage, pathological (T) stage, perineural and lymphovascular invasion, and lymph node metastasis were also noted. IHC was done for a panel of markers that was chosen on the basis of common markers used in the literature to classify intestinal and pancreatobiliary subtypes. These markers were CDX2, MUC2, and CK20 for intestinal subtype and MUC1, CK17, and CK7 for pancreatobiliary subtype. Normal intestinal mucosa was used as a positive control for MUC1, MUC2, CDX2, and CK20. Biliary epithelium in the tissue sections was taken as a positive internal control for CK7. Squamous
2.1. Statistics The χ2 test was used to correlate the positivity of different IHC markers with histologic parameters, and the KaplanMeier log-rank test was used to compare overall and median survival. Stepwise binary logistic regression was applied to test an independent association of IHC markers with histologic parameters, and Cox regression was applied to estimate the multivariate analysis between different IHC marker positivity and survival. Probability (P) values of less than .05 obtained by the 2-tailed significance tests were regarded as statistically significant. The SPSS software program, version 15 (version 15; SPSS Inc, Chicago, IL) was used for all statistical analyses.
3. Results
Intestinal and pancreatobiliary differentiation in PC
2215
Table 1 Histologic subtypes and associated findings in periampullary carcinoma according to anatomical location (n = 108) Histologic types
Ampulla CBD Duodenum Head of (n = 91) (n = 5) (n = 3) pancreas (n = 9)
Intestinal (n = 38) 35 Pancreatobiliary 43 (n = 53) Mixed (n = 8) 7 Others Adenosquamous 1 (n = 2) Undifferentiated (n 3 = 3) Signet ring cell 2 carcinoma (n = 4)
1 4
2 0
0 6
0
0
1
0
0
1
0
0
0
0
1
1
present in more than 70% of the area was considered the predominant type. The positivity of different IHC markers in the histologic subtypes of periampullary carcinoma is given in Table 2 (Fig. 1C).
3.1. Immunohistochemistry A panel of 6 markers were used to immunohistochemically classify the intestinal and pancreatobiliary subtypes of pancreatoduodenectomies of all locations taking morphology as criterion standard. CDX2 was the only useful marker with high sensitivity (89.5%) and specificity (100%) in differentiating intestinal from pancreatobiliary subtypes. MUC1 and CK7 were the markers with the least sensitivity and specificity (Table 3). In the present study, we found that nuclear stratification was not a specific feature of intestinaltype tumors. The cases that had nuclear stratification but the nuclei were round instead of elongated or cigar shaped and the location of nucleus was not basal but in the center of cell did not show positivity for CDX2. Such tumors were considered pancreatobiliary in the present study. Two cases of morphologic mixed type showed strong diffuse positivity for CDX2 but were negative for CK7 and CK17. Another 1 case of signet ring cell carcinoma showed positivity for both MUC2 and CK20 but was negative for CK7 and CK17. These 3 cases were reclassified as intestinal type. One case of morphologic undifferentiated type showed positivity for CK7, CK17, and MUC1 and was negative for all intestinal markers. Therefore, this case was reclassified as pancreatobiliary type after IHC. Thus, there was a change in the subtype of 4 cases (3.7%) in mixed, poorly differentiated, and undifferentiated tumors. The morphologic subtypes, IHC subtypes, and expression of IHC markers were correlated with different histologic variables and patients' survival. The morphologic and IHC subtypes showed a similar correlation with all variables being statistically significant. In CDX2-positive tumors,
Fig. 1 A, Intestinal-type periampullary carcinoma showing columnar tumor cells with elongated cigar-shaped basally located nuclei and nuclear stratification (hematoxylin and eosin stain, ×200 magnification). B, Pancreatobiliary-type periampullary carcinoma showing simple tubular glands with cuboidal to low columnar cells and a single layer of round centrally placed nuclei (hematoxylin and eosin stain, ×400 magnification). C, IHC showing strong positivity for CDX2 in intestinal-type periampullary carcinoma (hematoxylin and eosin stain, ×200 magnification).
2216 Table 2
N. Kumari et al. Expression of IHC markers in intestinal and pancreatobiliary subtypes of periampullary carcinoma (n = 108)
IHC markers
CDX2 (n = 43) MUC2 (n = 27) CK20 (n = 33) MUC1 (n = 108) CK17 (n = 23) CK7 (n = 90) CDX2+/MUC2+ (n = 21) CDX2+/CK20+ (n = 23) CK7+/CK17+ (n = 23) CDX2+/MUC2+/CK20+ (n = 12) MUC1+/CK17+/CK7+ (n = 23)
Histologic subtypes Intestinal (n = 38)
Pancreatobiliary (n = 53)
Mixed (n = 8)
Signet ring cell (n = 4)
Adenosquamous (n = 2)
Undifferentiated (n = 3)
34 (89.5%) 15 (39.5%) 19 (50%) 38 (100%) 4 (10.5%) 30 (78.9%) 13 (34.2%) 19 (50%) 4 (10.5%) 8 (21%) 4 (10.5%)
0 2 (3.7%) 7 (13.2%) 53 (100%) 17 (32.1%) 48 (90.5%) 0 0 17 (32.1%) 0 17 (32.1%)
5 6 4 8 1 6 5 2 1 2 1
2 (50%) 3 (75%) 2 (50%) 4 (100%) 0 2 (50%) 2 (50%) 1 (25%) 0 1 (25%) 0
1 0 0 2 1 2 0 0 1
1 (33.3%) 1 (33.3%) 1 (33.3%) 3 (100%) 0 2 (66.6%) 1 (33.3%) 1 (33.3%) 0 1 (33.3%) 0
except for tumor grade and lymph node metastasis, all other parameters were statistically significant (Table 4). CK17 expression showed a positive correlation with tumor grade and lymphovascular invasion. In the morphologic and IHC subtypes, the median survival was 45 and 44 months, respectively, in intestinal type and 20 months in pancreatobiliary type (P = .01 and P = .02). Of the panel of 6 IHC markers, CDX2 was the only marker associated with patients' survival and showed a median survival of 44 months in CDX2-positive tumors compared with 22 months in CDX2-negative tumors (P = .03, Kaplan-Meier log-rank test; Fig. 2). CDX2 expression was also independently associated with patients' survival on Cox regression analysis (P = .04). The rest of the 5 markers did not show any correlation with overall survival. The periampullary carcinomas in different anatomical locations (ampullary, head of pancreas, distal CBD, and duodenal) were analyzed stage-wise for overall survival, which showed that the ampullary tumors had the highest overall survival (median, 34 months) compared with tumors of distal CBD (median, 19 months) and head of pancreas (median, 10 months) in stage I tumors (P = .001). In stage II, the mean survival was 22, 29, and 12 months in ampullary, distal CBD, and pancreatic head tumors, respectively (P = .5). Table 3
(62.5%) (75%) (50%) (100%) (12.5%) (75%) (62.5%) (25%) (12.5%) (25%) (12.5%)
(50%)
(100%) (50%) (100%)
(50%)
1 (50%)
Similarly, in stage I tumors, the intestinal subtype had a higher overall survival (median, 41 months) than pancreatobiliary subtype (median, 19 months), with a P value of .009. In stage II, the median survival was 30 months in intestinal type and 20 months in pancreatobiliary type (P = .6). In CDX2-positive tumors, low–T stage (T1/T2) tumors had a median survival of 44 months as compared with high– T stage (T3/T4) tumors with 20 months (P = .01). The CDX2positive tumors showed an increased median of 48 months in patients without perineural invasion (PNI) compared with 24 months in CDX2-positive tumors with PNI (P = .001). The CDX2-positive tumors showed a median survival of 44 months in patients without lymphovascular invasion (LVI) compared with 11 months in patients with LVI (P = .06). There was no significant difference in survival within CDX2-positive tumors between present and absent lymph node metastasis (P = .9).
4. Discussion The histologic subtyping of periampullary carcinoma into intestinal and pancreatobiliary subtypes is emerging as one of the prognostic factors. Therefore, the methods used to classify these tumors should be more objective and reproducible.
Sensitivity and specificity of IHC markers in differentiating histologic subtypes of periampullary carcinoma
IHC markers
Sensitivity
Specificity
Positive predictive value
Negative predictive value
CDX2 MUC2 CK20 MUC1 CK17 CK7 CDX2+/MUC2+ CDX2+/CK20+ CK17+/CK7+ CDX2+/CK20+/MUC2+ MUC1+/CK17+/CK7+
89.5% 39.5% 50% 100% 32% 90.5% 34.2% 50% 32% 21% 32%
100% 96.2% 86.8% 0% 89.5% 21% 100% 100% 89.5% 100% 89.5%
100% 88.2% 73.1% 0 80.9% 61.5% 100% 100% 80.9% 100% 80.9%
93% 68.9% 70.8% 0 48.6% 61.5% 67.9% 73.6% 48.6% 63.8% 48.6%
Intestinal and pancreatobiliary differentiation in PC Table 4
2217
Correlation of histologic variables and median survival with morphologic and IHC subtypes and CDX2 expression
Histologic variables
Tumor grade 1 2 3 4 Tumor stage I II III Pathological (T) stage T1/T2 T3/T4 LN metastasis PNI LVI Median survival (mo)
Morphologic subtypes Intestinal (n = 38)
Pancreatobiliary (n = 53)
31 (81.5%) 6 (15.7%) 1 (2.6%) 0
27 (50.9%) 22 (41.5%) 4 (7.5%) 0
23 (60.6%) 14 (36.8%) 1 (2.6%)
15 (28.3%) 38 (71.7) 0
33 (86.8%) 5 (13.1%) 11 (28.9%) 7 (18.4%) 2 (5.2%) 45
31 (58.4%) 22 (41.5%) 28 (52.8%) 32 (60.3%) 16 (30.1%) 20
P
IHC subtypes
P
Intestinal (n = 41)
Pancreatobiliary (n = 54)
32 (78.1%) 7 (17.1%) 2 (4.8%) 0
27 (50%) 22 (40.7%) 4 (7.4%) 1 (1.9%)
25 (61%) 15 (36.6%) 1 (2.4%)
15 (27.8%) 39 (72.2%) 0
36 (87.8%) 5 (12.2%) 12 (29.2%) 9 (21.9%) 3 (7.3%) 44
31 (57.4%) 23 (42.6%) 28 (51.8%) 33 (61.1%) 16 (29.6%) 20
.01
P
Positive (n = 43)
Negative (n = 65)
31 (72.1%) 8 (18.6%) 4 (9.3%) 0
34 (52.3%) 23 (35.4%) 6 (9.2%) 2 (3%)
23 (53.5%) 18 (41.9%) 2 (4.6%)
21 (32.3%) 44 (67.7%) 0
35 (81.4%) 8 (18.6%) 14 (32.5%) 11 (25.6%) 4 (9.3%) 44
41 24 33 36 17 22
.04
.03
.1
.002
.04
.02 .001 .003 .01
CD2 expression
.01
.001
Westgaard et al [8] found a moderate agreement between classification using morphologic criteria and IHC classification with a κ coefficient of 0.53, but they had an almost-perfect interobserver agreement in subtyping by morphologic criteria alone with a κ coefficient of 0.9. In the present study, we classified 108 cases of periampullary carcinoma using morphologic criteria and IHC markers into intestinal, pancreatobiliary, mixed, and other subtypes and found an almost-perfect interobserver agreement between morphologic classification with a κ of 0.84 and between morphologic and IHC classification with a κ of 0.94.This difference would be because of different panels of antibodies and variable cutoffs used for the IHC markers used in both studies. In the present study, of the panel of 6 antibodies done to differentiate intestinal and pancreatobiliary subtypes, CDX2 was the only marker that had both high sensitivity (89.5%) and high specificity (100%) for predicting intestinal subtype (considering 10% staining of tumor cells as positive). The second marker that had a moderate sensitivity of 50% and a specificity of 86.8% was CK20. The other 4 markers had either low sensitivity or low specificity or both (Table 3). MUC1 was positive in all the 108 cases and did not help in subtyping periampullary carcinomas in the present study. The cutoff for positive staining for different markers has been variable in different studies ranging from any cell positive to as high as 25% to 30%. Therefore, the sensitivity in the literature ranges from 54% to 100% for CDX2, 78% to 83% for CK17, 13% to 74% for MUC2, and 49% to 100% for MUC1 in different studies [4,5,8–10]. The specificity as reported in the literature is 70% to 85% for CDX2, 37% to 81% for CK17, 76% to 100% for MUC2, and 46.5% to 81.8% for MUC1, in different studies [4,5,8–10]. This shows that CDX2 has a narrow but high range of sensitivity and specificity as compared with CK17, MUC2, and MUC1 in different studies.
.02 .001 .007 .02
.04 (63.1%) (36.9%) (50.8%) (55.4%) (26.1%)
.08 .002 .03 .03
Westgaard et al [8] considered any tumor cell positive as positive staining for CDX2 and at least 10% tumor cells positive for other markers used. They found CDX2 to be positive in 54.3% intestinal type and 14.9% pancreatobiliary subtype, with a sensitivity of 54% and specificity of 85%.They found CK7 positivity (P = .009), CDX2 negativity (P = .002), and MUC4 positivity (P = .026) to be independent markers in their study to predict pancreatobiliary subtype. Sessa et al [10] used MUC1, MUC2, MUC5AC, MUC6, and CDX2 and found 100% sensitivity and 70% specificity of CDX2 for intestinal differentiation, considering 10% cutoff for positive staining. de Paiva Haddad et al [9] used CK7, CK17, CK20, MUC1, MUC2, MUC5AC, MUC6, CDX2, and CD10 for subtyping periampullary carcinoma and considered a positive staining as more than 5% positive tumor cells. The markers that showed significantly higher frequency of positivity for intestinal type were MUC2, CK20, CD10, and CDX2, and for pancreatobiliary type, they were CK7 and MUC1. CDX2 expression was seen in 86% intestinal type and 21.3% pancreatobiliary type, with 86% sensitivity and 78.7% specificity. In their study, CK17 showed 78.7% sensitivity and 37.2% specificity for pancreatobiliary differentiation. de Paiva Haddad et al had also good agreement between IHC and histologic classification (κ = 0.702) but did not find a significant association of CDX2 expression with outcome. This may be because of a lower cutoff (5%) used for CDX2 and a comparable number of intestinal- and pancreatobiliary-type cancers. Chu et al [4] found 100% sensitivity and 83.3% specificity of CDX2 for intestinal type, whereas 83.3% sensitivity and 81.8% specificity of CK17 for pancreatobiliary type, considering positive staining when at least more than 5% tumor cells were positive. Kawabata et al [5] used a combination of CK20 and MUC1 and considered at least 10% tumor cells positive as positive staining in their study.
2218
Fig. 2 Overall survival curve showing better survival in CDX2positive tumors (Kaplan-Meier log-rank test, P = .03).
They found CK20+/MUC1− expression in 100% intestinal type and CK20−/MUC1+ expression in 94% pancreatobiliary type and revised their 9 cases of unusual type into 2 cases of pancreatobiliary and 1 case of intestinal type after IHC. Thus, the present study demonstrates high CDX2 sensitivity and specificity similar to other studies at a uniform 10% cutoff for positive staining, as is considered most IHC markers. The wide range of sensitivity and specificity for all the IHC markers could be attributed to different cohorts of cases studied (ampullary only, pancreatic head only, or periampullary as a whole), variable cutoffs considered as positive staining, and use of different clones of antibodies. The IHC markers that are used as an adjunct to subtype periampullary carcinoma have also shown correlation with prognosis. The histologic prognostic parameters and patients' survival were correlated with expression of IHC markers where CDX2 positive tumors showed a statistically significant low tumor stage, more T1/T2 stage tumors, and less perineural and lymphovascular invasion (Table 4). The next marker that showed some correlation was CK17. CK17positive tumors had a high tumor grade and more lymphovascular invasion compared with CK17-negative tumors; however, it showed a low sensitivity of 32% (Table 3). In the multivariate analysis using stepwise binary logistic regression, CDX2 showed a negative correlation with perineural invasion, with a P value of .004. Westgaard et al [8] found that cancers expressing MUC1 and/or MUC4 indicated poor prognosis and were associated with pancreatobiliary type of differentiation. Kitamura et al [11] found that ampullary carcinoma with stronger MUC1 stains and weaker MUC2 staining had worse prognosis. Hansel et al [12] demonstrated expression of CDX2 as an independent marker of outcome in periampullary carcinoma. The correlation of histologic prognostic parameters and patients' survival with morphologic and IHC subtypes and
N. Kumari et al. CDX2 expression (Table 4) showed that intestinal and pancreatobiliary subtypes behaved similarly when classified either by morphologic criteria or after combining morphology with IHC markers. When compared with CDX2 expression, all parameters, except for tumor grade and lymph node metastasis, were largely similar. This may be caused by the analysis of all cases of CDX2 positive and negative, irrespective of histologic differentiation where some cases of mixed and other types also showed CDX2 expression. Thus, the present study shows that morphologic criteria alone are sufficient for histologic subtyping and using a panel of multiple IHC markers was helpful in only a limited number of cases (3.7%). Thus, IHC has some role in subtyping where it can be used in difficult cases of mixed or other types of tumors. Westgaard et al, in their recent study, emphasized the histologic differentiation of the periampullary group of tumors and showed that the correlation of histologic prognostic parameters in a particular subtype (intestinal or pancreatobiliary) was similar irrespective of anatomical location whether ampullary, pancreatic head, distal CBD, or duodenal cancer [13]. In the present study, CDX2 was the only IHC marker associated with patients' overall survival in periampullary carcinoma in both Kaplan-Meier log-rank test and Cox regression analysis. The median survival in CDX2-positive tumors was 44 months compared with 22 months, with a P value of .03, which was similar to that shown by Hansel et al [12]. They showed that CDX2-positive cancers had a median survival of 45 months versus 20 months for CDX2negative cancers. The 5-year survival rate in their study was 76% in CDX2-positive cancers and 26% in CDX2-negative cancers [12]. Westgaard et al [13] also showed an increased median survival in intestinal than pancreatobiliary type (60 months versus 17 months) irrespective of anatomical location. We also compared the median overall survival in tumors classified by morphologic and IHC criteria and CDX2 expression and found that survival was significantly longer in intestinal-type tumors classified by morphology alone or by using morphology and IHC. Similar findings were seen with CDX2-positive tumors. In this study, we also showed that within CDX2-positive tumors, cases with T1/T2 tumors, no PNI, and no LVI had longer survival compared with T3/T4 tumors or tumors with PNI and LVI. We correlated the overall survival with different anatomical subsets and histologic differentiation in different tumor stages and found that ampullary location and intestinal morphology had higher survival in stage I but not in stage II tumors. This may be explained by fewer cases presenting in anatomical locations other than ampullary, and also, only few cases in these locations showed intestinal differentiation.
5. Conclusion We have shown that using panel of IHC markers is not helpful in differentiating intestinal and pancreatobiliary
Intestinal and pancreatobiliary differentiation in PC subtypes. CDX2 is helpful, to some extent, in histologic differentiation in difficult cases and also predicting longer survival in CDX2-positive cases similar to that of intestinal subtype. In summary, we have shown that intestinal and pancreatobiliary subtypes of periampullary carcinoma can be classified satisfactorily on morphology alone. IHC helped in classifying another 3.7% cases that were difficult on routine histology. Thus, a panel of IHC markers may be essential only in tumors of mixed or other subtypes that are difficult on morphology. CDX2 expression has the highest sensitivity and specificity for intestinal-type differentiation and also has an added advantage of predicting longer overall survival in the tumors of the periampullary group, whether of intestinal or nonintestinal morphology. Thus, we suggest that CDX2 may be included routinely in the subtyping of periampullary carcinomas as a whole, irrespective of anatomical location of histologic type. This may be of some help in the future if some kinds of adjuvant therapies are developed for particular histologic subtypes.
References [1] Westgaard A, Tafjord S, Farstad IN, et al. Pancreatobiliary versus intestinal histologic type of differentiation is an independent prognostic factor in resected periampullary adenocarcinoma. BMC Cancer 2008;8:170. [2] Albores-Saavedra J. Tumors of the gallbladder, extrahepatic bile ducts, and ampulla of Vater. In: Albores-Saavedra J, editor. Atlas of tumor pathology. Washington, DC: Armed Forces Institute of Pathology; 2000. p. 316.
2219 [3] Zhou H, Schaefer N, Wolff M, Fischer HP. Carcinoma of the ampulla of Vater: comparative histologic/immunohistochemical classification and follow up. Am J Surg Pathol 2004;28:875-82. [4] Chu PG, Schwarz RE, Lau SK, Yen Y, Weiss LM. Immunohistochemical staining in the diagnosis of pancreatobiliary and ampulla of Vater adenocarcinoma: application of CDX2, CK17, MUC1, and MUC2. Am J Surg Pathol 2005;29:359-67. [5] Kawabata Y, Tanaka T, Nishisaka T, Inao T, Nishi T, Yano S. Cytokeratin 20 (CK20) and apomucin 1 (MUC1) expression in ampullary carcinoma: correlation with tumor progression and prognosis. Diagn Pathol 2010;5:75. [6] Fischer HP, Zhou H. Pathogenesis of carcinoma of the papilla of Vater. J Hepatobiliary Pancreat Surg 2004;11:301-9. [7] Paulsen FP, Varoga D, Paulsen AR, Corfield A, Tsokos M. Prognostic value of mucins in the classification of ampullary carcinomas. HUM PATHOL 2006;37:160-7. [8] Westgaard A, Schjølberg AR, Cvancarova M, Eide TJ, Clausen OP, Gladhaug IP. Differentiation markers in pancreatic head adenocarcinomas: MUC1 and MUC4 expression indicates poor prognosis in pancreatobiliary differentiated tumors. Histopathology 2009;54:337-47. [9] de Paiva Haddad LB, Patzina RA, Penteado S, et al. Lymph node involvement and not the histopathologic subtype are correlated with outcome after resection of adenocarcinoma of the ampulla of Vater. J Gastrointest Surg 2010;14:719-28. [10] Sessa F, Furlan D, Zampatti C, Carnevali I, Franzi F, Capella C. Prognostic factors for ampullary adenocarcinomas: tumor stage, tumor histology, tumor location, immunohistochemistry and microsatellite instability. Virchows Arch 2007;451:649-57. [11] Kitamura H, Yonezawa S, Tanaka S, Kim YS, Sato E. Expression of mucin carbohydrates and core proteins in carcinomas of the ampulla of Vater: their relationship to prognosis. Jpn J Cancer Res 1996;87:631-40. [12] Hansel DE, Maitra A, Lin JW, et al. Expression of the caudal-type homeodomain transcription factors CDX 1/2 and outcome in carcinomas of the ampulla of Vater. J Clin Oncol 2005;23:1811-8. [13] Westgaard A, Pomianowska E, Clausen OPF, Gladhaug IP. Intestinaltype and pancreatobiliary-type adenocarcinomas: how does ampullary carcinoma differ from other periampullary malignancies? Ann Surg Oncol 2013;20:430-9.
www.elsevier.com/locate/humpath
Original contribution
Intestinal and pancreatobiliary differentiation in periampullary carcinoma: the role of immunohistochemistry☆ Niraj Kumari MD, DNB, MNAMS a,⁎, Kanchan Prabha MD a , Rajneesh Kumar Singh MS, MCh b , Dinesh Kumar Baitha MSc a , Narendra Krishnani MD a a
Department of Pathology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow 266014, India Department of Surgical Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow 266014, India b
Received 5 February 2013; revised 26 April 2013; accepted 1 May 2013
Keywords: Periampullary carcinoma; Intestinal; Pancreatobiliary; CDX2; Ampullary carcinoma
Summary Periampullary carcinoma (PC) is classified into intestinal and pancreatobiliary subtypes using morphology and immunohistochemistry (IHC). Different combinations of markers have been used in the literature. One hundred eight PCs were classified using morphology and IHC (CDX2, mucin [MUC] 2, cytokeratin [CK] 20, CK7, CK17, and MUC1). The expression of these markers was compared with different histologic subtypes, histopathologic prognostic parameters, and patients' survival. There were 38 intestinal and 53 pancreatobiliary subtypes classified on morphology alone. CDX2 showed high sensitivity (89.5%) and specificity (100%) for intestinal type. CK20 and MUC2 showed low sensitivity (50% and 39.5%) but high specificity (86.8% and 96.2%) for intestinal type. CK7 and CK17 showed a sensitivity of 90.5% and 32% and a specificity of 21% and 89.4%, respectively, for pancreatobiliary subtype. MUC1 was 100% sensitive but 0% specific in pancreatobiliary subtype. The overall median survival in morphologic and IHC intestinal type was 45 months versus 20 months in pancreatobiliary type (P = 0.01). Intestinal and pancreatobiliary types of PC were differentiated in 84.2% of cases by morphology alone and in 87.9% cases with IHC. CDX2-positive tumors had a median survival of 44 months versus 22 months in CDX2-negative tumors (P = .03). IHC helped in reclassifying an additional 4 cases of mixed and other types. Among the panel used, CDX2 showed a high sensitivity and specificity for intestinal subtype and was an independent prognostic marker for longer survival. Thus, CDX2 may be used routinely with morphology in subtyping of PC, and a panel of markers may be used in morphologically difficult cases. © 2013 Elsevier Inc. All rights reserved.
1. Introduction ☆
Funding: The work was funded by the SGPGI Intramural Research Committee of the institute. ⁎ Corresponding author. E-mail addresses: [email protected] (N. Kumari), [email protected] (K. Prabha), [email protected] (R. K. Singh), [email protected] (D. K. Baitha), [email protected] (N. Krishnani). 0046-8177/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.humpath.2013.05.003
The histologic subtyping of periampullary carcinoma into intestinal and pancreatobiliary subtypes is emerging as an important prognostic factor. Despite the common location, 2 different histologic subtypes are associated with different premalignant lesion, cell-type specific markers, oncogene expression, mode of cancer spread, extent of cancers, and
2214
N. Kumari et al.
interaction with extracellular matrix. Westgaard et al [1] suggested that difference in survival of periampullary carcinoma of comparable size is more dependent on histologic differentiation than anatomical origin. The subtyping of periampullary carcinoma into intestinal and pancreatobiliary types was proposed by Albores-Saavedra [2]. He proposed the morphologic criteria that defined intestinal type as cancers composed of well-formed tubular glands, complex cribriform areas, and solid nests, which are indistinguishable from colorectal adenocarcinoma. The tumor cells are pseudostratified tall columnar with oval- or cigar-shaped nuclei located basally and often contain mucin (MUC). The pancreatobiliary type was defined as cancers composed of simple or branching glands and small solid nests of cells surrounded by dense desmoplastic stroma. The tumor cells are cuboidal to low columnar arranged in a single layer without nuclear pseudostratification, and nuclei are rounded with marked variation in size and shape from one cell to the other. Several studies have classified periampullary carcinoma subtypes using the above morphologic criteria alone or in combination of a panel of immunohistochemistry (IHC) markers. The different markers that have been used in the literature to classify these 2 subtypes include cytokeratins (CK7, CK17, and CK20), MUCs (MUC1, MUC2, MUC4, and MUC5AC), and transcription factor CDX2 [3–7]. CK7, CK17, MUC4, and MUC1 have been found to be expressed in pancreatobiliary type of periampullary carcinomas, whereas CK20, MUC2, and CDX2 are expressed in intestinal-type cancers [5]. Paulsen et al [7] used 8 different types of MUCs including MUC1, MUC2, MUC4, MUC5AC, MUC5B, MUC6, MUC7, and MUC8 and found that MUC2, MUC7, and MUC8 were helpful in classifying ampullary carcinoma. The present study aimed to analyze the usefulness of IHC as single or panel of markers that included different CKs, MUCs, and transcription factor CDX2 that would add to the morphologic criteria in subtyping of periampullary carcinoma.
cell carcinoma was used as a positive control for CK17. The antibodies for MUC1 (polyclonal, dilution 1:50) and CK17 (clone E3, dilution 1:40) were obtained from Thermoscientific (San Jose, CA, USA). MUC2 antibody (clone Ccp58, dilution 1:100) was procured from Novocastra (Newcastle, UK), CDX2 antibody (MU392A.UC, dilution 1:50) was procured from Biogenex (Fremond, CA, USA). CK7 (clone OV-TL 12/30, dilution 1:80), and CK20 antibody (clone Ks 20.8, dilution 1:80) was procured from Dako (Glostrup, Denmark). Briefly, the sections were cut at 4-μm thickness and deparaffinized in xylene and graded alcohol. Antigen retrieval was done by microwave treatment using EZ Antigen Retriever (Biogenex) in citrate buffer at pH 6.0 for MUC1, MUC2, and CK17 and in EDTA buffer at pH 8.0 for CDX2, CK7, and CK20 for 30 minutes. Primary antibody was added for a 2-hour duration at room temperature, followed by secondary antibody for 30 minutes (DAKO Real Envision). The slides were then washed and counterstained with Mayer hematoxylin. Positive staining was considered when at least 10% or more tumor cells were stained.
2. Materials and methods
There were 74 men and 34 women (male/female, 2.1:1), with an age range of 27 to 81 years (mean, 57.2 years; median, 58 years). Anatomically, there were 91 ampullary carcinomas, 9 carcinomas of the head of pancreas, 5 distal common bile duct (CBD) carcinomas, and 3 duodenal carcinomas. The histologic subtypes of periampullary carcinoma in different anatomical subsets are given in Table 1. There were 38 (35.1%) cases of intestinal type (Fig. 1A), 53 (49%) cases of pancreatobiliary type (Fig. 1B), 8 (7.4%) cases of mixed type, and 9 (8.3%) cases of other types classified on the basis of morphologic criteria. Mixedtype tumors are classified based on the predominant histologic differentiation, and a defined cutoff has not been proposed; however, in the present study, tumors were classified as mixed based on an arbitrary cutoff of 70:30 ratios, where the intestinal or pancreatobiliary component ranged from 30% to 70%. Any histologic subtype when
One hundred eight consecutive cases of pancreatoduodenectomies were reviewed to classify them as intestinal, pancreatobiliary, mixed, or others subtypes based on the morphologic criteria given by Albores-Saavedra [2]. The different histologic parameters including tumor grade, stage, pathological (T) stage, perineural and lymphovascular invasion, and lymph node metastasis were also noted. IHC was done for a panel of markers that was chosen on the basis of common markers used in the literature to classify intestinal and pancreatobiliary subtypes. These markers were CDX2, MUC2, and CK20 for intestinal subtype and MUC1, CK17, and CK7 for pancreatobiliary subtype. Normal intestinal mucosa was used as a positive control for MUC1, MUC2, CDX2, and CK20. Biliary epithelium in the tissue sections was taken as a positive internal control for CK7. Squamous
2.1. Statistics The χ2 test was used to correlate the positivity of different IHC markers with histologic parameters, and the KaplanMeier log-rank test was used to compare overall and median survival. Stepwise binary logistic regression was applied to test an independent association of IHC markers with histologic parameters, and Cox regression was applied to estimate the multivariate analysis between different IHC marker positivity and survival. Probability (P) values of less than .05 obtained by the 2-tailed significance tests were regarded as statistically significant. The SPSS software program, version 15 (version 15; SPSS Inc, Chicago, IL) was used for all statistical analyses.
3. Results
Intestinal and pancreatobiliary differentiation in PC
2215
Table 1 Histologic subtypes and associated findings in periampullary carcinoma according to anatomical location (n = 108) Histologic types
Ampulla CBD Duodenum Head of (n = 91) (n = 5) (n = 3) pancreas (n = 9)
Intestinal (n = 38) 35 Pancreatobiliary 43 (n = 53) Mixed (n = 8) 7 Others Adenosquamous 1 (n = 2) Undifferentiated (n 3 = 3) Signet ring cell 2 carcinoma (n = 4)
1 4
2 0
0 6
0
0
1
0
0
1
0
0
0
0
1
1
present in more than 70% of the area was considered the predominant type. The positivity of different IHC markers in the histologic subtypes of periampullary carcinoma is given in Table 2 (Fig. 1C).
3.1. Immunohistochemistry A panel of 6 markers were used to immunohistochemically classify the intestinal and pancreatobiliary subtypes of pancreatoduodenectomies of all locations taking morphology as criterion standard. CDX2 was the only useful marker with high sensitivity (89.5%) and specificity (100%) in differentiating intestinal from pancreatobiliary subtypes. MUC1 and CK7 were the markers with the least sensitivity and specificity (Table 3). In the present study, we found that nuclear stratification was not a specific feature of intestinaltype tumors. The cases that had nuclear stratification but the nuclei were round instead of elongated or cigar shaped and the location of nucleus was not basal but in the center of cell did not show positivity for CDX2. Such tumors were considered pancreatobiliary in the present study. Two cases of morphologic mixed type showed strong diffuse positivity for CDX2 but were negative for CK7 and CK17. Another 1 case of signet ring cell carcinoma showed positivity for both MUC2 and CK20 but was negative for CK7 and CK17. These 3 cases were reclassified as intestinal type. One case of morphologic undifferentiated type showed positivity for CK7, CK17, and MUC1 and was negative for all intestinal markers. Therefore, this case was reclassified as pancreatobiliary type after IHC. Thus, there was a change in the subtype of 4 cases (3.7%) in mixed, poorly differentiated, and undifferentiated tumors. The morphologic subtypes, IHC subtypes, and expression of IHC markers were correlated with different histologic variables and patients' survival. The morphologic and IHC subtypes showed a similar correlation with all variables being statistically significant. In CDX2-positive tumors,
Fig. 1 A, Intestinal-type periampullary carcinoma showing columnar tumor cells with elongated cigar-shaped basally located nuclei and nuclear stratification (hematoxylin and eosin stain, ×200 magnification). B, Pancreatobiliary-type periampullary carcinoma showing simple tubular glands with cuboidal to low columnar cells and a single layer of round centrally placed nuclei (hematoxylin and eosin stain, ×400 magnification). C, IHC showing strong positivity for CDX2 in intestinal-type periampullary carcinoma (hematoxylin and eosin stain, ×200 magnification).
2216 Table 2
N. Kumari et al. Expression of IHC markers in intestinal and pancreatobiliary subtypes of periampullary carcinoma (n = 108)
IHC markers
CDX2 (n = 43) MUC2 (n = 27) CK20 (n = 33) MUC1 (n = 108) CK17 (n = 23) CK7 (n = 90) CDX2+/MUC2+ (n = 21) CDX2+/CK20+ (n = 23) CK7+/CK17+ (n = 23) CDX2+/MUC2+/CK20+ (n = 12) MUC1+/CK17+/CK7+ (n = 23)
Histologic subtypes Intestinal (n = 38)
Pancreatobiliary (n = 53)
Mixed (n = 8)
Signet ring cell (n = 4)
Adenosquamous (n = 2)
Undifferentiated (n = 3)
34 (89.5%) 15 (39.5%) 19 (50%) 38 (100%) 4 (10.5%) 30 (78.9%) 13 (34.2%) 19 (50%) 4 (10.5%) 8 (21%) 4 (10.5%)
0 2 (3.7%) 7 (13.2%) 53 (100%) 17 (32.1%) 48 (90.5%) 0 0 17 (32.1%) 0 17 (32.1%)
5 6 4 8 1 6 5 2 1 2 1
2 (50%) 3 (75%) 2 (50%) 4 (100%) 0 2 (50%) 2 (50%) 1 (25%) 0 1 (25%) 0
1 0 0 2 1 2 0 0 1
1 (33.3%) 1 (33.3%) 1 (33.3%) 3 (100%) 0 2 (66.6%) 1 (33.3%) 1 (33.3%) 0 1 (33.3%) 0
except for tumor grade and lymph node metastasis, all other parameters were statistically significant (Table 4). CK17 expression showed a positive correlation with tumor grade and lymphovascular invasion. In the morphologic and IHC subtypes, the median survival was 45 and 44 months, respectively, in intestinal type and 20 months in pancreatobiliary type (P = .01 and P = .02). Of the panel of 6 IHC markers, CDX2 was the only marker associated with patients' survival and showed a median survival of 44 months in CDX2-positive tumors compared with 22 months in CDX2-negative tumors (P = .03, Kaplan-Meier log-rank test; Fig. 2). CDX2 expression was also independently associated with patients' survival on Cox regression analysis (P = .04). The rest of the 5 markers did not show any correlation with overall survival. The periampullary carcinomas in different anatomical locations (ampullary, head of pancreas, distal CBD, and duodenal) were analyzed stage-wise for overall survival, which showed that the ampullary tumors had the highest overall survival (median, 34 months) compared with tumors of distal CBD (median, 19 months) and head of pancreas (median, 10 months) in stage I tumors (P = .001). In stage II, the mean survival was 22, 29, and 12 months in ampullary, distal CBD, and pancreatic head tumors, respectively (P = .5). Table 3
(62.5%) (75%) (50%) (100%) (12.5%) (75%) (62.5%) (25%) (12.5%) (25%) (12.5%)
(50%)
(100%) (50%) (100%)
(50%)
1 (50%)
Similarly, in stage I tumors, the intestinal subtype had a higher overall survival (median, 41 months) than pancreatobiliary subtype (median, 19 months), with a P value of .009. In stage II, the median survival was 30 months in intestinal type and 20 months in pancreatobiliary type (P = .6). In CDX2-positive tumors, low–T stage (T1/T2) tumors had a median survival of 44 months as compared with high– T stage (T3/T4) tumors with 20 months (P = .01). The CDX2positive tumors showed an increased median of 48 months in patients without perineural invasion (PNI) compared with 24 months in CDX2-positive tumors with PNI (P = .001). The CDX2-positive tumors showed a median survival of 44 months in patients without lymphovascular invasion (LVI) compared with 11 months in patients with LVI (P = .06). There was no significant difference in survival within CDX2-positive tumors between present and absent lymph node metastasis (P = .9).
4. Discussion The histologic subtyping of periampullary carcinoma into intestinal and pancreatobiliary subtypes is emerging as one of the prognostic factors. Therefore, the methods used to classify these tumors should be more objective and reproducible.
Sensitivity and specificity of IHC markers in differentiating histologic subtypes of periampullary carcinoma
IHC markers
Sensitivity
Specificity
Positive predictive value
Negative predictive value
CDX2 MUC2 CK20 MUC1 CK17 CK7 CDX2+/MUC2+ CDX2+/CK20+ CK17+/CK7+ CDX2+/CK20+/MUC2+ MUC1+/CK17+/CK7+
89.5% 39.5% 50% 100% 32% 90.5% 34.2% 50% 32% 21% 32%
100% 96.2% 86.8% 0% 89.5% 21% 100% 100% 89.5% 100% 89.5%
100% 88.2% 73.1% 0 80.9% 61.5% 100% 100% 80.9% 100% 80.9%
93% 68.9% 70.8% 0 48.6% 61.5% 67.9% 73.6% 48.6% 63.8% 48.6%
Intestinal and pancreatobiliary differentiation in PC Table 4
2217
Correlation of histologic variables and median survival with morphologic and IHC subtypes and CDX2 expression
Histologic variables
Tumor grade 1 2 3 4 Tumor stage I II III Pathological (T) stage T1/T2 T3/T4 LN metastasis PNI LVI Median survival (mo)
Morphologic subtypes Intestinal (n = 38)
Pancreatobiliary (n = 53)
31 (81.5%) 6 (15.7%) 1 (2.6%) 0
27 (50.9%) 22 (41.5%) 4 (7.5%) 0
23 (60.6%) 14 (36.8%) 1 (2.6%)
15 (28.3%) 38 (71.7) 0
33 (86.8%) 5 (13.1%) 11 (28.9%) 7 (18.4%) 2 (5.2%) 45
31 (58.4%) 22 (41.5%) 28 (52.8%) 32 (60.3%) 16 (30.1%) 20
P
IHC subtypes
P
Intestinal (n = 41)
Pancreatobiliary (n = 54)
32 (78.1%) 7 (17.1%) 2 (4.8%) 0
27 (50%) 22 (40.7%) 4 (7.4%) 1 (1.9%)
25 (61%) 15 (36.6%) 1 (2.4%)
15 (27.8%) 39 (72.2%) 0
36 (87.8%) 5 (12.2%) 12 (29.2%) 9 (21.9%) 3 (7.3%) 44
31 (57.4%) 23 (42.6%) 28 (51.8%) 33 (61.1%) 16 (29.6%) 20
.01
P
Positive (n = 43)
Negative (n = 65)
31 (72.1%) 8 (18.6%) 4 (9.3%) 0
34 (52.3%) 23 (35.4%) 6 (9.2%) 2 (3%)
23 (53.5%) 18 (41.9%) 2 (4.6%)
21 (32.3%) 44 (67.7%) 0
35 (81.4%) 8 (18.6%) 14 (32.5%) 11 (25.6%) 4 (9.3%) 44
41 24 33 36 17 22
.04
.03
.1
.002
.04
.02 .001 .003 .01
CD2 expression
.01
.001
Westgaard et al [8] found a moderate agreement between classification using morphologic criteria and IHC classification with a κ coefficient of 0.53, but they had an almost-perfect interobserver agreement in subtyping by morphologic criteria alone with a κ coefficient of 0.9. In the present study, we classified 108 cases of periampullary carcinoma using morphologic criteria and IHC markers into intestinal, pancreatobiliary, mixed, and other subtypes and found an almost-perfect interobserver agreement between morphologic classification with a κ of 0.84 and between morphologic and IHC classification with a κ of 0.94.This difference would be because of different panels of antibodies and variable cutoffs used for the IHC markers used in both studies. In the present study, of the panel of 6 antibodies done to differentiate intestinal and pancreatobiliary subtypes, CDX2 was the only marker that had both high sensitivity (89.5%) and high specificity (100%) for predicting intestinal subtype (considering 10% staining of tumor cells as positive). The second marker that had a moderate sensitivity of 50% and a specificity of 86.8% was CK20. The other 4 markers had either low sensitivity or low specificity or both (Table 3). MUC1 was positive in all the 108 cases and did not help in subtyping periampullary carcinomas in the present study. The cutoff for positive staining for different markers has been variable in different studies ranging from any cell positive to as high as 25% to 30%. Therefore, the sensitivity in the literature ranges from 54% to 100% for CDX2, 78% to 83% for CK17, 13% to 74% for MUC2, and 49% to 100% for MUC1 in different studies [4,5,8–10]. The specificity as reported in the literature is 70% to 85% for CDX2, 37% to 81% for CK17, 76% to 100% for MUC2, and 46.5% to 81.8% for MUC1, in different studies [4,5,8–10]. This shows that CDX2 has a narrow but high range of sensitivity and specificity as compared with CK17, MUC2, and MUC1 in different studies.
.02 .001 .007 .02
.04 (63.1%) (36.9%) (50.8%) (55.4%) (26.1%)
.08 .002 .03 .03
Westgaard et al [8] considered any tumor cell positive as positive staining for CDX2 and at least 10% tumor cells positive for other markers used. They found CDX2 to be positive in 54.3% intestinal type and 14.9% pancreatobiliary subtype, with a sensitivity of 54% and specificity of 85%.They found CK7 positivity (P = .009), CDX2 negativity (P = .002), and MUC4 positivity (P = .026) to be independent markers in their study to predict pancreatobiliary subtype. Sessa et al [10] used MUC1, MUC2, MUC5AC, MUC6, and CDX2 and found 100% sensitivity and 70% specificity of CDX2 for intestinal differentiation, considering 10% cutoff for positive staining. de Paiva Haddad et al [9] used CK7, CK17, CK20, MUC1, MUC2, MUC5AC, MUC6, CDX2, and CD10 for subtyping periampullary carcinoma and considered a positive staining as more than 5% positive tumor cells. The markers that showed significantly higher frequency of positivity for intestinal type were MUC2, CK20, CD10, and CDX2, and for pancreatobiliary type, they were CK7 and MUC1. CDX2 expression was seen in 86% intestinal type and 21.3% pancreatobiliary type, with 86% sensitivity and 78.7% specificity. In their study, CK17 showed 78.7% sensitivity and 37.2% specificity for pancreatobiliary differentiation. de Paiva Haddad et al had also good agreement between IHC and histologic classification (κ = 0.702) but did not find a significant association of CDX2 expression with outcome. This may be because of a lower cutoff (5%) used for CDX2 and a comparable number of intestinal- and pancreatobiliary-type cancers. Chu et al [4] found 100% sensitivity and 83.3% specificity of CDX2 for intestinal type, whereas 83.3% sensitivity and 81.8% specificity of CK17 for pancreatobiliary type, considering positive staining when at least more than 5% tumor cells were positive. Kawabata et al [5] used a combination of CK20 and MUC1 and considered at least 10% tumor cells positive as positive staining in their study.
2218
Fig. 2 Overall survival curve showing better survival in CDX2positive tumors (Kaplan-Meier log-rank test, P = .03).
They found CK20+/MUC1− expression in 100% intestinal type and CK20−/MUC1+ expression in 94% pancreatobiliary type and revised their 9 cases of unusual type into 2 cases of pancreatobiliary and 1 case of intestinal type after IHC. Thus, the present study demonstrates high CDX2 sensitivity and specificity similar to other studies at a uniform 10% cutoff for positive staining, as is considered most IHC markers. The wide range of sensitivity and specificity for all the IHC markers could be attributed to different cohorts of cases studied (ampullary only, pancreatic head only, or periampullary as a whole), variable cutoffs considered as positive staining, and use of different clones of antibodies. The IHC markers that are used as an adjunct to subtype periampullary carcinoma have also shown correlation with prognosis. The histologic prognostic parameters and patients' survival were correlated with expression of IHC markers where CDX2 positive tumors showed a statistically significant low tumor stage, more T1/T2 stage tumors, and less perineural and lymphovascular invasion (Table 4). The next marker that showed some correlation was CK17. CK17positive tumors had a high tumor grade and more lymphovascular invasion compared with CK17-negative tumors; however, it showed a low sensitivity of 32% (Table 3). In the multivariate analysis using stepwise binary logistic regression, CDX2 showed a negative correlation with perineural invasion, with a P value of .004. Westgaard et al [8] found that cancers expressing MUC1 and/or MUC4 indicated poor prognosis and were associated with pancreatobiliary type of differentiation. Kitamura et al [11] found that ampullary carcinoma with stronger MUC1 stains and weaker MUC2 staining had worse prognosis. Hansel et al [12] demonstrated expression of CDX2 as an independent marker of outcome in periampullary carcinoma. The correlation of histologic prognostic parameters and patients' survival with morphologic and IHC subtypes and
N. Kumari et al. CDX2 expression (Table 4) showed that intestinal and pancreatobiliary subtypes behaved similarly when classified either by morphologic criteria or after combining morphology with IHC markers. When compared with CDX2 expression, all parameters, except for tumor grade and lymph node metastasis, were largely similar. This may be caused by the analysis of all cases of CDX2 positive and negative, irrespective of histologic differentiation where some cases of mixed and other types also showed CDX2 expression. Thus, the present study shows that morphologic criteria alone are sufficient for histologic subtyping and using a panel of multiple IHC markers was helpful in only a limited number of cases (3.7%). Thus, IHC has some role in subtyping where it can be used in difficult cases of mixed or other types of tumors. Westgaard et al, in their recent study, emphasized the histologic differentiation of the periampullary group of tumors and showed that the correlation of histologic prognostic parameters in a particular subtype (intestinal or pancreatobiliary) was similar irrespective of anatomical location whether ampullary, pancreatic head, distal CBD, or duodenal cancer [13]. In the present study, CDX2 was the only IHC marker associated with patients' overall survival in periampullary carcinoma in both Kaplan-Meier log-rank test and Cox regression analysis. The median survival in CDX2-positive tumors was 44 months compared with 22 months, with a P value of .03, which was similar to that shown by Hansel et al [12]. They showed that CDX2-positive cancers had a median survival of 45 months versus 20 months for CDX2negative cancers. The 5-year survival rate in their study was 76% in CDX2-positive cancers and 26% in CDX2-negative cancers [12]. Westgaard et al [13] also showed an increased median survival in intestinal than pancreatobiliary type (60 months versus 17 months) irrespective of anatomical location. We also compared the median overall survival in tumors classified by morphologic and IHC criteria and CDX2 expression and found that survival was significantly longer in intestinal-type tumors classified by morphology alone or by using morphology and IHC. Similar findings were seen with CDX2-positive tumors. In this study, we also showed that within CDX2-positive tumors, cases with T1/T2 tumors, no PNI, and no LVI had longer survival compared with T3/T4 tumors or tumors with PNI and LVI. We correlated the overall survival with different anatomical subsets and histologic differentiation in different tumor stages and found that ampullary location and intestinal morphology had higher survival in stage I but not in stage II tumors. This may be explained by fewer cases presenting in anatomical locations other than ampullary, and also, only few cases in these locations showed intestinal differentiation.
5. Conclusion We have shown that using panel of IHC markers is not helpful in differentiating intestinal and pancreatobiliary
Intestinal and pancreatobiliary differentiation in PC subtypes. CDX2 is helpful, to some extent, in histologic differentiation in difficult cases and also predicting longer survival in CDX2-positive cases similar to that of intestinal subtype. In summary, we have shown that intestinal and pancreatobiliary subtypes of periampullary carcinoma can be classified satisfactorily on morphology alone. IHC helped in classifying another 3.7% cases that were difficult on routine histology. Thus, a panel of IHC markers may be essential only in tumors of mixed or other subtypes that are difficult on morphology. CDX2 expression has the highest sensitivity and specificity for intestinal-type differentiation and also has an added advantage of predicting longer overall survival in the tumors of the periampullary group, whether of intestinal or nonintestinal morphology. Thus, we suggest that CDX2 may be included routinely in the subtyping of periampullary carcinomas as a whole, irrespective of anatomical location of histologic type. This may be of some help in the future if some kinds of adjuvant therapies are developed for particular histologic subtypes.
References [1] Westgaard A, Tafjord S, Farstad IN, et al. Pancreatobiliary versus intestinal histologic type of differentiation is an independent prognostic factor in resected periampullary adenocarcinoma. BMC Cancer 2008;8:170. [2] Albores-Saavedra J. Tumors of the gallbladder, extrahepatic bile ducts, and ampulla of Vater. In: Albores-Saavedra J, editor. Atlas of tumor pathology. Washington, DC: Armed Forces Institute of Pathology; 2000. p. 316.
2219 [3] Zhou H, Schaefer N, Wolff M, Fischer HP. Carcinoma of the ampulla of Vater: comparative histologic/immunohistochemical classification and follow up. Am J Surg Pathol 2004;28:875-82. [4] Chu PG, Schwarz RE, Lau SK, Yen Y, Weiss LM. Immunohistochemical staining in the diagnosis of pancreatobiliary and ampulla of Vater adenocarcinoma: application of CDX2, CK17, MUC1, and MUC2. Am J Surg Pathol 2005;29:359-67. [5] Kawabata Y, Tanaka T, Nishisaka T, Inao T, Nishi T, Yano S. Cytokeratin 20 (CK20) and apomucin 1 (MUC1) expression in ampullary carcinoma: correlation with tumor progression and prognosis. Diagn Pathol 2010;5:75. [6] Fischer HP, Zhou H. Pathogenesis of carcinoma of the papilla of Vater. J Hepatobiliary Pancreat Surg 2004;11:301-9. [7] Paulsen FP, Varoga D, Paulsen AR, Corfield A, Tsokos M. Prognostic value of mucins in the classification of ampullary carcinomas. HUM PATHOL 2006;37:160-7. [8] Westgaard A, Schjølberg AR, Cvancarova M, Eide TJ, Clausen OP, Gladhaug IP. Differentiation markers in pancreatic head adenocarcinomas: MUC1 and MUC4 expression indicates poor prognosis in pancreatobiliary differentiated tumors. Histopathology 2009;54:337-47. [9] de Paiva Haddad LB, Patzina RA, Penteado S, et al. Lymph node involvement and not the histopathologic subtype are correlated with outcome after resection of adenocarcinoma of the ampulla of Vater. J Gastrointest Surg 2010;14:719-28. [10] Sessa F, Furlan D, Zampatti C, Carnevali I, Franzi F, Capella C. Prognostic factors for ampullary adenocarcinomas: tumor stage, tumor histology, tumor location, immunohistochemistry and microsatellite instability. Virchows Arch 2007;451:649-57. [11] Kitamura H, Yonezawa S, Tanaka S, Kim YS, Sato E. Expression of mucin carbohydrates and core proteins in carcinomas of the ampulla of Vater: their relationship to prognosis. Jpn J Cancer Res 1996;87:631-40. [12] Hansel DE, Maitra A, Lin JW, et al. Expression of the caudal-type homeodomain transcription factors CDX 1/2 and outcome in carcinomas of the ampulla of Vater. J Clin Oncol 2005;23:1811-8. [13] Westgaard A, Pomianowska E, Clausen OPF, Gladhaug IP. Intestinaltype and pancreatobiliary-type adenocarcinomas: how does ampullary carcinoma differ from other periampullary malignancies? Ann Surg Oncol 2013;20:430-9.