OUP user menu

CDX-2 Expression Is a Common Event in Primary Intestinal-type Endocervical Adenocarcinoma

Reda S. Saad MD, PhD, Nadia Ismiil MD, Valérie Dubé MD, Sharon Nofech-Mozes MD, Mahmoud A. Khalifa MD, PhD
DOI: http://dx.doi.org/10.1309/AJCP7E5ASGOENPFP 531-538 First published online: 1 October 2009


We studied the expression of cytokeratin (CK) 7, CK20, CDX-2, and p16 in 119 cervical adenocarcinomas (65 usual type [50 invasive; 15 in situ], 37 intestinal type [21 invasive; 16 in situ], 10 endometrioid, 5 adenosquamous, and 2 signet-ring carcinomas) in comparison with 55 cases of rectal adenocarcinomas. The percentage of cells staining was considered negative if 0% to 5% stained; more than 5% was considered positive. For p16, staining of more than 50% was considered positive. CK7 was expressed in all cervical cases and in 12 rectal adenocarcinomas (22%). CK20 was expressed in 17 cervical adenocarcinomas (14.3%) and in 48 rectal adenocarcinomas (87%). CK20 immunostaining was diffuse in the majority of rectal tumors but focal in most cervical tumors. CDX-2 was expressed in all cases of rectal adenocarcinoma and in 46 cervical adenocarcinomas (38.7%): usual type, 10 (15%); intestinal type, 31 (84%); endometrioid type, 5 (50%); adenosquamous and signet-ring types, 0 (0%). CDX-2 is a marker for intestinal differentiation irrespective of a rectal or cervical origin. Therefore, it should not be used as the sole basis to confirm the colorectum as the primary origin in metastatic cases.

Key Words:
  • CDX-2
  • Cervical adenocarcinoma
  • Intestinal differentiation

Metastatic cancer of unknown origin accounts for approximately 3% to 5% of all malignant neoplasms.1 Identifying the primary site of a metastatic tumor can lead to more effective treatment and improve the overall outcome. Despite recent advances in imaging technology, the origin of metastatic tumor remains unknown on many occasions.2 Therefore, immunohistochemical staining patterns have a crucial diagnostic role in characterizing the site of origin of these tumors.3

Cytokeratin (CK) 7 and CK20 are low-molecular-weight cytokeratins that can be useful in identifying the origin of metastatic adenocarcinomas.39 In addition, transcription factors such as CDX-2 have been detected in adult tissues from the duodenum to the rectum.10,11 Studies have shown that CDX-2 is a sensitive and a specific marker for the identification of colorectal carcinoma.1214 However, CDX-2 expression was also found in other carcinomas such as upper gastrointestinal tract, ovarian mucinous, endometrioid, urinary bladder tumors, and other carcinomas with intestinal-type morphologic features.12,13,15,16 The relationship between CDX-2 expression and intestinal differentiation suggests that CDX-2 immunoreactivity is not specific for neoplasms of the intestinal tract but may be indicative of intestinal differentiation within an adenocarcinoma.17

Intestinal differentiation of cervical adenocarcinoma, in the form of goblet cells and/or Paneth cells, is uncommon but may generate a diagnostic dilemma. Invasive cervical adenocarcinoma with intestinal differentiation could mimic the histologic features of colorectal adenocarcinoma, raising the possibility of metastasis or direct spread, especially in a small cervical biopsy specimen. Also, a distant metastasis from an intestinal-type cervical adenocarcinoma could be easily mistaken for a metastatic adenocarcinoma of intestinal origin on a morphologic basis alone. Clinical and radiologic examination may not be definitive in ascertaining where the tumor arose. Because management of cervical and intestinal adenocarcinomas is fundamentally different, the correct identification of the origin could affect patient treatment. Therefore, immunohistochemical analysis might be helpful in distinguishing between the different types of tumors, as enteric markers expression in a cervical neoplasm support the diagnosis of metastatic colorectal adenocarcinoma. However, the immunophenotype of intestinal-type cervical adenocarcinoma has not been fully investigated. The purpose of this study was to investigate the immunophenotype of different types of cervical adenocarcinoma, with emphasis on the intestinal type, in comparison with primary rectal adenocarcinoma.

Materials and Methods

Case Selection

Cervical adenocarcinoma specimens were retrieved from the surgical pathology files of the Department of Pathology, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, Canada, between January 2000 and December 2007. We collected a total of 119 consecutive cervical adenocarcinoma cases: 65 endocervical/usual type (50 invasive and 15 in situ), 37 intestinal type (21 invasive and 16 in situ), 10 endometrioid type, 5 adenosquamous, and 2 signet-ring carcinomas. No history of concurrent adenocarcinoma elsewhere in the body was noted in any of these patients. In addition, 55 cases of rectal adenocarcinoma were included in the study as a control group.

The original slides were reviewed by 2 pathologists (R.S.S. and N.I.) in all cases to confirm the original diagnosis of primary cervical and rectal adenocarcinoma. Histologic subtyping of the neoplasms was assigned according to the World Health Organization classification of tumors.18 The endocervical variant (usual-type) exhibited stratified cells with abundant eosinophilic cytoplasm and pale endocervical-type mucin, and the endometrioid variant resembled tumors found in the endometrium. Intestinal differentiation was diagnosed in the presence of goblet cells and/or Paneth cells. In all cases of intestinal-type cervical adenocarcinoma, foci of usual type were also observed. Only the immunophenotype of the intestinal-type component was recorded for this category. Discrepancies were resolved by consensus at a multiheaded microscope.

Tissue samples from the specimens were fixed in 10% buffered formalin, processed, and stained for H&E routinely according to the local procedural guidelines in our laboratory. One paraffin block with the maximum amount of tumor and proper fixation was chosen from immunohistochemical studies. The study was approved by the Sunnybrook Health Sciences Centre Research Ethics Board.

Immunohistochemical Analysis

For the study, 4-μm tissue sections were deparaffinized and rehydrated in a descending series of ethanol solutions. All cases were immunostained for CK7, CK20, CDX-2, and p16. Table 1 summarizes the characteristics, pretreatment, dilutions, incubation periods, localization, and sources of the antibodies. Tissue sections were mounted on coated slides and dried for 1 hour at 60°C.

After preincubation with 1% hydrogen peroxide for 10 minutes to block the endogenous peroxidase activity, the antibodies were incubated at appropriate dilutions for the period and temperature indicated in Table 1. An automated stainer (Ventana Immunostainer, Ventana Medical Systems, Tucson, AZ) and a basic diaminobenzidine immunohistochemical detection system (Ventana Medical Systems) were used for the rest of the procedure. Sections were counterstained with Mayer hematoxylin, dehydrated, cleared in xylene, and mounted. Sections of benign colonic mucosa served as positive control samples for CDX-2 and CK20, lung tumor for CK7, and cervical carcinoma for p16. For negative control samples, the primary antibody was omitted with each run.

Microscopic Evaluation

Two pathologists (R.S.S. and N.I.) assessed the immunostaining without knowledge of the previous clinical, radiologic, or histopathologic diagnoses. For CDX-2, only nuclear staining was considered positive. Cytoplasmic positivity was infrequently encountered and was considered nonspecific. Cytoplasmic staining with CK7 and CK20 was considered positive. For p16, concurrent nuclear and cytoplasmic staining was required for a case to be reported as positive. The percentage of positive cells was recorded as follows: 0, 0% to 5%; 1+, 6% to 25%; 2+, 26% to 50%; 3+, 51% to 75%; and 4+, more than 75%. Cases that stained 0% to 5% were considered negative and all others as positive for statistical purposes, except for p16, for which cases staining more than 50% were considered as positive. After the cases were evaluated, diagnoses were revealed and staining patterns of the various tumors compared. Interobserver variability was resolved by consensus for the slide in question under a multiheaded microscope.

View this table:
Table 1

Statistical Analysis

Statistical analysis was performed to determine the sensitivity, specificity, positive predictive value, and negative predictive value.


The median age at diagnosis for endocervical adenocarcinoma patients was 45 years (mean ± SD, 35 ± 49 years). In the 119 cases, total abdominal hysterectomy was performed in 64 (53.8%), cone or loop electrode excision procedure biopsy in 49 (41.2%), and cervical biopsy in 6 (5.0%).

View this table:
Table 2
View this table:
Table 3

Table 2 shows the percentage of cases that stained positively for CK7, CK20, and CDX-2 in different types of cervical adenocarcinoma and rectal adenocarcinoma. Table 3 shows the sensitivity, specificity, positive predictive value, and negative predictive value for CK7, CK20, p16, and CDX-2 immunoreactivity and of CK7+/CK20− and CK7−/CK20+ immunostaining patterns as diagnostic markers for rectal vs cervical adenocarcinoma.

CK7 and CK20

CK20 was expressed in 48 (87%) of 55 cases of rectal adenocarcinoma and in only 17 (14.3%) of 119 cases of cervical adenocarcinoma. In the different types of cervical adenocarcinomas, CK20 expression was as follows: 4 (6%) of 65 usual type (1/15 in situ; 3/50 invasive), 8 (22%) of 37 intestinal type (3/16 in situ; 5/21 invasive) Image 1 , 3 (30%) of 10 endometrioid adenocarcinomas, 0 of 5 adenosquamous carcinomas, and 2 (100%) of 2 cases of signet-ring carcinoma. CK7 was expressed in all cervical adenocarcinoma cases (100%), while it was expressed in only 12 (22%) of 55 rectal tumors.

The CK7−/CK20+ immunophenotype was expressed by 36 (65%) of 55 rectal tumors and was not observed in any of the endocervical tumors. The CK7+/CK20+ immunophenotype was expressed in 12 (22%) of 55 rectal adenocarcinomas and in 17 (14.3%) of 119 cervical adenocarcinomas. However, CK20 reactivity was diffuse (staining of >50% of cells) in the majority of rectal carcinomas (42/55 [76%]) but only focal (<50% of cells positive) in 10 (59%) of 17 cases of cervical adenocarcinoma. The CK7+/CK20− immunophenotype was not expressed in any of the rectal adenocarcinomas, whereas it was expressed in 102 (85.7%) of 119 cervical adenocarcinomas. Of 55 rectal adenocarcinomas, 7 (13%) showed a CK7−/CK20− immunophenotype (each marker staining <5% of the cells), whereas none of the cervical tumors showed this pattern.


CDX-2 was expressed in all cases of rectal adenocarcinoma. In positive cases, the immunoreactivity was predominantly nuclear with occasional faint cytoplasmic staining. The majority of cases (53/55 [96%]) demonstrated strong and diffuse immunostaining in more than 50% of cells.

Overall, the nuclear staining for CDX-2 was expressed in 46 (38.7%) of 119 cases of cervical adenocarcinoma. Nuclear immunopositivity was more frequent in the intestinal type, which showed expression in 31 (84%) of 37 cases (in situ, 12/16 [75%]; invasive, 19/21 [90%]) Image 2 , followed by the endometrioid subtype, which was positive in 5 (50%) of 10 cases Image 3A . CDX-2 was expressed in 10 (15%) of 65 usual cervical adenocarcinoma cases (in situ, 2/15 [13%]; invasive, 8/50 [16%]) Image 3B . None of the adenosquamous or signet-ring carcinomas expressed CDX-2. The pattern of staining in some cases was notable for discrete clusters of malignant glands with nuclear staining and areas of intervening negative glands, whereas other cases had scattered positive cells throughout the tumor.

In the usual endocervical type, all CDX-2 immunoreactive cases demonstrated focal staining (<50% cells staining). In comparison, intestinal-type cervical adenocarcinomas showed diffuse staining for CDX-2 (>50% of cells positive) in the majority of invasive cases (12/21 [57%]) but only in a minority of in situ cases (2/16 [13%]).


In 7 (13%) of 55 rectal adenocarcinoma cases, p16 was expressed diffusely (>50% staining cells) Image 4 . Focal staining was observed in 5 (9%) of 55 as 1+ and 8 (15%) of 55 as 2+. In positive cases, p16 showed cytoplasmic and nuclear staining. In cervical adenocarcinoma, p16 was detected in 105 (88.2%) of 119 cases, including 59 (91%) of 65 of the usual type, 32 (86%) of 37 of the intestinal type, 7 (70%) of 10 of the endometrioid type, 5 (100%) of 5 of the adenosquamous type, and 1 (50%) of 2 of the signet-ring type.

Image 1

Cytokeratin 20 expression in intestinal-type cervical adenocarcinoma (A) and endometrioid-type cervical adenocarcinoma (B) (A and B, ×400).


There has been little study of CK7, CK20, and CDX-2 immunoreactivity in the different types of cervical adenocarcinoma.19,20 An adenocarcinoma usually exhibits a characteristic cytokeratin profile, similar to that of its native epithelium and that can be used to identify the origin of a metastatic tumor.2023 A CK7+/CK20− immunophenotype is characteristic of müllerian-type epithelia, whereas a CK7−/CK20+ pattern is considered specific for colorectal epithelial tumors. Similar to previous studies, our results illustrate that different types of cervical adenocarcinoma diffusely express CK7. Similar to ovarian mucinous carcinoma,16,2428 CK20 was focally positive in a minority of cases of cervical adenocarcinoma (17/119 [14.3%]). A recent study showed that CK20 was always negative in adenocarcinoma in situ of the usual type but positive in the majority of invasive adenocarcinomas (5/6 cases).29 Similarly, in our study, CK20 was positive in more invasive than in situ cervical adenocarcinomas. Intestinal-type endocervical adenocarcinomas typically express p16, similar to their usual-type counterparts, correlating with high-risk human papillomavirus (HPV) infection.2932

Image 2

A, CDX-2 staining in intestinal-type endocervical adenocarcinoma in situ. Note the negative staining in normal part of the gland and positive staining in the dysplastic component (×400). B, Invasive carcinoma (CDX-2, ×400).

Image 3

CDX-2 staining in endocervical adenocarcinoma in the endometrioid type (A) and the usual type (B) (A and B, ×400).

Our results confirm that rectal carcinoma expressed CK7 in 22% of cases compared with 16% reported by Chu et al6 and 13% reported by Ramalingam et al.33 In contrast, a recent study reported a high figure for CK7 positivity in rectal carcinoma, up to 74%.34 The reason for this discrepancy is unknown but may result from differences in the studied population or difficulty in separating adenocarcinomas arising in the anorectal region (usually CK7+/CK20+) from conventional rectal adenocarcinomas, or it may indicate the heterogeneity of rectal carcinoma.35,36

Image 4

Expression of p16 in rectal adenocarcinoma (×400).

CDX-2 is a nuclear transcription factor that has previously been demonstrated to have a critical role in the pathogenesis of colorectal adenocarcinomas.1214 By using surgical and cytologic material, Barbareschi et al14 and Saad et al37 found similar results, with CDX-2 expression limited to primary and metastatic colorectal adenocarcinomas compared with the negative staining of primary lung adenocarcinomas. Previous studies have shown that CDX-2 protein expression could be used to identify colorectal adenocarcinomas with great sensitivity.1214,38 However, expression of CDX-2 in tumors other than colorectal carcinoma has been previously reported. Werling et al12 and Moskaluk et al13 reported that a significant fraction of gastric adenocarcinomas, ovarian mucinous carcinomas, and primary bladder adenocarcinomas are CDX-2+. In gynecologic tumors, CDX-2 has been reported to be positive in 0% to 100% of ovarian mucinous tumors and in 0% to 30% of ovarian endometrioid carcinomas, whereas serous tumors have only been rarely positive.16,25,27,28,39

CDX-2 immunostaining has been rarely studied in large series of cervical adenocarcinomas of various histologic subtypes, and the available results are rather conflicting. Tornillo et al26 reported a metastatic cervical adenocarcinoma that was CDX-2−. Raspollini et al19 reported 2 cases of intestinal-type cervical adenocarcinomas, and both were CDX-2−, similar to the findings of Sullivan et al.17 In contrast, McCluggage et al29 recently reported CDX-2 positivity in the majority of cases of intestinal-type endocervical adenocarcinoma in situ (20/21) and in all 3 invasive intestinal-type adenocarcinomas studied. However, most previous studies analyzed only a few cases of invasive intestinal-type cervical adenocarcinoma. Our study is the largest reported to date and confirms that the majority of invasive and in situ adenocarcinomas of the intestinal type show CDX-2 immunoreactivity. In this regard our results are in agreement with those of McCluggage et al.29

For cervical adenocarcinomas of the usual type, Sullivan et al17 reported CDX-2 positivity in 7 (33%) of 21 cases they studied and McCluggage et al29 in 1 (17%) of 6 cases. In our study, 8 (16%) of 50 cases of invasive usual-type adenocarcinoma were positive for CDX-2. The higher percentage described by Sullivan et al17 can possibly be explained by the fact that they used a 1% cutoff to consider a case as positive, whereas we used a 5% cutoff. In endometrioid adenocarcinomas, Sullivan et al17 reported 2 of 3 of their cases positive for CDX-2, similar to our results (50%). Tot40 found that a CK7−/CK20+ expression pattern was more specific for colonic adenocarcinoma metastases than CDX-2 alone (95.3% vs 90%) but less sensitive (79.5% vs 84%).

The majority of cervical adenocarcinomas are associated with high-risk HPV. Previous reports showed that HPV in situ hybridization testing is a specific marker for a cervical origin of metastatic adenocarcinomas, with a sensitivity ranging from 78% to 91%.4143 In addition, HPV was negative in colorectal adenocarcinomas.44 These results suggest that HPV DNA testing could be a useful marker to distinguish between endocervical adenocarcinoma and metastatic colorectal carcinoma, particularly in cases in which a management decision is crucial.

The majority of cervical adenocarcinomas of the intestinal type express CDX-2, whereas they are usually negative or only focally positive for CK20. CDX-2 is also expressed in other histologic types of cervical adenocarcinomas to a lesser extent. Therefore, CDX-2 is not totally specific for colorectal carcinoma but may be a marker of intestinal differentiation. The CK20+/CK7− immunophenotype is a more specific marker for colorectal origin than CDX-2 expression alone. An immunohistochemical panel combining CK7, CK20, and p16 is of value in distinguishing a primary cervical intestinal-type adenocarcinoma from a colorectal adenocarcinoma. This may be of value in locally advanced cases of primary cervical or colorectal adenocarcinoma in which the clinical, radiologic, and morphologic features cannot definitively determine the site of origin or in the investigative workup of a distant metastasis of unknown origin.


We thank Mei Xu, a university of Toronto student, LMP research summer program, for help in collecting the material included in this study.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
View Abstract