Materials and methods

Tissue Microarray Construction

A tissue microarray of 1420 unselected, non-consecutive colorectal cancer was constructed as described previously.³⁴ Formalin-fixed, paraffin-embedded tissue blocks of colorectal cancer resections were retrieved from the archives of the Institute of Pathology, University Hospital of Basel, Switzerland, the Institute of Clinical Pathology, Basel Switzerland and the Institute of Pathology, Stadtspital Triemli, Zürich, Switzerland. One-tissue cylinder with a diameter of 0.6 mm was punched from morphologically representative tissue areas of each 'donor' tissue block and brought into one-recipient paraffin block (3 2.5 cm) using a homemade semiautomated tissue arrayer. Failure of analysis (<10%) was related to tissue microarray technology including missing samples or fractions containing only a few tumor cells.

Normal Colon Tissue

In addition to colorectal cancer, 57 normal colon tissues were integrated into the tissue microarray and were used as a control group to compare expression of protein markers.

Immunohistochemistry of Tissue Microarray

Immunohistochemistry was performed for protein markers MLH1, MSH2, MSH6, CD8, CK20, CK7 and CDX2. Briefly, four-micrometer sections of tissue microarray blocks were transferred to an adhesive-coated slide system (Instrumedics Inc., Hackensack, NJ, USA) to facilitate the transfer of tissue microarray sections to slides and to minimize tissue loss. Standard indirect immunoperoxidase procedures were used for immunohistochemistry (ABC-Elite, Vector Laboratories, Burlingame, CA, USA). All 1420 colorectal cancer and 57 normal colonic mucosa samples were immunostained for CK20 (clone Ks20.8, 1:50, DAKO, Baar, Switzerland), CK7 (clone OV-TL 12/30, dilution 1:200, DAKO, Baar, Switzerland), CDX2 (clone AMT28, 1:50, Abcam, Cambridge, UK), MLH1 (clone MLH-1; dilution 1:100; BD Biosciences Pharmingen, San Jose, CA, USA), MSH2 (clone MSH-2; dilution 1:200; BD Biosciences Pharmingen, San Jose, CA, USA), MSH6 (clone 44; dilution 1:400; BD Biosciences Pharmingen, San Jose, CA, USA) and CD8 (clone C8/144B, dilution 1:100; Dako Cytomation, Carpinteria, USA). After dewaxing and rehydration in dH₂O, sections for immunostaining were subjected to heat antigen retrieval in a microwave oven (1200 W, 15 min) in 0.001 mol/l ethylenediaminetetraacetic acid, pH 8.0 for MLH1 and MSH2, pH 9 for CK20 and CK7 and in 0.01 mol/l citrate buffer, pH 6.0 for MSH6 and pH 2 for CDX2. Endogenous peroxidase activity was blocked using 0.5% H₂O₂. After transfer to a humidified chamber, the sections were incubated with 10% normal goat serum (Dako Cytomation) for 20 min and incubated with primary antibody at room temperature for 1 h (CK20, CK7 and CDX2). Subsequently, the sections were incubated with peroxidase-labeled polymer (K4005, EnVision+ System-HRP(AEC); DakoCytomation) for 30 min at room temperature. For visualization of the antigen, the sections were immersed in 3-amino-9-ethylcarbazole+substrate-chromogen (K4005, EnVision+ System-HRP (AEC); DakoCytomation) for 30 min, and counterstained lightly with Gill's hematoxylin.

Mismatch Repair Status and Clinico-Pathological Data

Immunohistochemistry was used to identify mismatch repair-proficient and mismatch repair-deficient cases. For a tumor to be considered mismatch repair-proficient, positive expression (>0% tumor cell staining) for MLH1, MSH2 and MSH6 was required. All 1197 cases were determined to be mismatch repair-proficient. The remaining 223 tumors were defined as microsatellite instability high and showed absence of expression of either nuclear MLH1 or MSH2 or MSH6. These microsatellite instability high cases included both sporadic MLH1-negative tumors (n=141) and 82 presumed HNPCC cases with loss of MSH2 and/or MSH6 at any age or loss of MLH1 at <55 years.³⁵ In addition, clinico-pathological data for all patients included age, tumor diameter, tumor location and complete follow-up as well as pT and pN stage, tumor grade and the presence of vascular invasion which were all assessed pathologically from the corresponding surgical resection specimen. In addition, disease-specific survival time was available for 1410 patients.

Evaluation of Immunohistochemistry

All cases were scored by one experienced pathologist (AL). Cases where the interpretation of the immunohistochemical analysis was difficult were discussed with a second experienced pathologist (LT). Cytoplasmic and/or membranous CK20 and CK7 and nuclear CDX2 were scored as the number of positive tumor cells. To consider a case-positive standard assessment of >0% of tumor cells was assigned when analyzing the diagnostic role of these markers and multi-marker phenotypes. To evaluate their prognostic effect and role in tumor progression, cutoff scores for positivity were optimized using receiver operating characteristic (ROC) curve analysis. Intraepithelial CD8+ intratumoral lymphocytes located in direct contact with tumor cells were quantified over the area of the entire punch for each case in the tissue microarray. Tumor in at least 50% of the tissue microarray punch was necessary for the case to be considered evaluable. Staining intensity was not evaluated.³⁶

Statistical Analysis

Differences in mean CK20, CK7 and CDX2 protein expression between mismatch repair-proficient and -deficient colorectal cancers were obtained by the Wilcoxon test. To determine the association of positive protein expression with clinico-pathological features, cutoff scores for CK20, CK7, and CDX2 expression were determined by ROC curve analysis.³⁷ The ROC curve is a plot of the sensitivity and (1-specificity) for an outcome at each value of the protein expression score. A protein score can therefore be selected from the curve such that a cutoff at this value leads to the greatest number of patients correctly classified as with (maximizing sensitivity) and without (maximizing specificity) the clinical end point. The (0, 1)-criterion was used to minimize the trade-off between sensitivity and specificity. To determine the reliability of the ROC curve-derived cutoff scores for CK20, CK7 and CDX2, 100 bootstrapped replications of the data were performed to re-sample the data. The most frequently obtained cutoff score was selected as the final score above which tumors should be considered positive for the outcome. ROC curve analysis and bootstrapping was repeated for each clinico-pathological end point. The ² and Fisher's Exact Test were used to determine the association between negative and positive CK20 and CK7 expression with T stage, N stage, tumor grade, vascular invasion and tumor location, whereas the Kaplan–Meier method and log-rank test were used for univariate survival analysis. Cox proportional hazards regression was carried out for multivariable survival analysis. Hazard ratios (HR) and 95% confidence intervals (CI) were obtained. P-values <0.05 were considered statistically significant. All analyses were carried out using SAS V.9 (The SAS Institute, Cary, NC, USA).

Results

Mismatch Repair Status

Mismatch repair-deficient tumors compared with mismatch repair-proficient cases displayed the typical characteristics such as preponderance for females (65.0 vs 50.1%; P<0.001), more frequently located in the right side (65.0 vs 29.1%; P<0.001) and more poorly differentiated (22.3 vs 11.0%; P<0.001) (Table 1). In addition, both 5- and 10-year survival rates are significantly improved in mismatch repair-deficient colorectal cancer compared with their mismatch repair-proficient counterparts (P<0.001). These findings thus confirm the representativeness of mismatch repair-deficiency and -proficiency in our patient cohort. Median follow-up time was 46.0 months.

Table 1 - Characteristics of patients with mismatch repair-proficient and -deficient colorectal cancer.

Full table

Normal Colonic Mucosa

In normal colonic mucosa CX20, CK7 and CDX2 expression could be detected in 98.4, 0 and 100% of cases, respectively.

Frequency of CK20, CK7 and CDX2 Expression in Colorectal Cancer Stratified by Mismatch Repair Status

The mean percentage of positive tumor cells expressing CK20 and CDX2 was higher in mismatch repair-proficient compared with mismatch repair-deficient colorectal cancer (P<0.001) (Table 2). There was no statistically significant difference in CK7 expression between mismatch repair-proficient and mismatch repair-deficient colorectal cancer (P=0.743). In addition, tumors were analyzed according to localization within the colorectum. Rectal cancers (n=482) were significantly more frequently positive for CK20 (P<0.001) and CDX2 (P<0.001) but not for CK7 (P=0.987).

Table 2 - Mean percentage of positive tumor cells expressing CK20, CK7 and CDX2 in mismatch repair-proficient and -deficient colorectal cancer.

Full table

CK20/7 Phenotypes in Mismatch Repair-Proficient and -Deficient Colorectal Cancer

In mismatch repair-deficient colorectal cancer CK20/7 phenotypes including CK20 negativity (CK20-/CK7- and CK20-/CK7+) were more frequent than in mismatch repair-proficient colorectal cancers (19.3% vs 7.5%) (Figure 1) (Table 3). The typical phenotype for colorectal cancer (CK20+/CK7-) was found in 84.5% of mismatch repair-proficient and in 73.1% of mismatch repair-deficient colorectal cancer.

Figure 1.

Percentage of CDX2- and CK20-negative cases is higher in mismatch repair-deficient (19.3 and 21.6%) compared with mismatch repair-proficient colorectal cancer (6.7 and 7.5%).

Full figure and legend (37K)

Table 3 - Distribution of CK20 and CK7 expression in mismatch repair-proficient and -deficient colorectal cancer.

Full table

CK20/7/CDX2 Phenotypes In Mismatch Repair-Proficient And Deficient Colorectal Cancer

In mismatch repair-deficient colorectal cancer CK20/7/CDX2 phenotypes including CDX2 negativity (CK20-/CK7-/CDX2-, CK20-/CK7+/CDX2-, CK20+/CK7-/CDX2- and CK20+/CK7+/CDX2-) could be more frequently detected compared with mismatch repair-proficient colorectal cancer (21.6. vs 6.7%) (Figure 1) (Table 4).

Table 4 - Distribution of CK20, CK7 and CDX2 expression in mismatch repair-proficient and -deficient colorectal cancer.

Full table

Association of CK20, CK7 and CDX2 with Clinico-Pathological Parameters

Loss of CK20 was associated with higher tumor grade in mismatch repair-proficient and -deficient colorectal cancer (P<0.001) and with location (right side) in mismatch repair-deficient colorectal cancer (P=0.02). In mismatch repair-proficient colorectal cancer there was an association between loss of CDX2 and higher T stage (P<0.001), N stage (P<0.001), tumor grade (P<0.001), vascular invasion (P=0.007) and location (right side) (<0.001), whereas in mismatch repair-deficient colorectal cancer CDX2 loss was associated with location (right side) (P=0.013) (Tables 5, 6, 7).

Table 5 - Association of CK20 with clinico-pathological features in mismatch repair-proficient and -deficient colorectal cancer.

Full table

Table 6 - Association of CK7 with clinico-pathological features in mismatch repair-proficient and -deficient colorectal cancer.

Full table

Table 7 - Association of CDX2 with clinico-pathological features in mismatch repair-proficient and -deficient colorectal cancer.

Full table

Multivariable Analysis

In mismatch repair-proficient colorectal cancer CK20 expression was an independent adverse prognostic factor adjusting for T and N, grade, vascular invasion, age and tumor side (left vs right) (P-value=0.002; HR (95% CI)=1.39 (1.14–1.69)), whereas CDX2 (P=0.958) and CK7 (P=0.129) were not independent prognostic factors.

Survival Analysis of CK20 Expression in Mismatch Repair-Proficient Colorectal Cancer Stratified by Intratumoral Lymphocytes

To further investigate the prognostic role of CK20 in mismatch repair-proficient colorectal cancer, the association between CK20 expression and intratumoral lymphocytes was analyzed. CK20 positivity was associated with absence of CD8+ intratumoral lymphocytes in both mismatch repair-deficient (P<0.001) and mismatch repair-proficient (P=0.02) tumors. Additionally, the stratification of the CK20 analysis by intratumoral lymphocytes showed a better survival in subgroups with presence of intratumoral lymphocytes independently of CK20 expression. Indeed, CK20-positive colorectal cancers with presence of intratumoral lymphocytes had a better survival than CK20-negative colorectal cancer without intratumoral lymphocytes (P<0.001) (Figure 2).

Figure 2.

Kaplan–Meier survival curve in mismatch repair-proficient colorectal cancer of (1) CK20- with presence of intratumoral lymphocytes, (2) CK20+ with presence of intratumoral lymphocytes, (3) CK20- with absence of intratumoral lymphocytes, (4) CK20+ with absence of intratumoral lymphocytes (P< 0.001).

Full figure and legend (54K)

Discussion

According to the revised Bethesda guidelines,³⁸ several pathologic features are associated with microsatellite instability status including the presence of intratumoral lymphocytes, a Crohn-like lymphocytic reaction, mucinous or signet ring differentiation, and a medullary or undifferentiated growth pattern.^{24, 25, 26, 27, 28, 29, 30, 31, 32} Jenkins et al³⁹ recently proposed a score to predict microsatellite instability high colorectal cancer based on characteristic features associated with this subgroup of tumors including age at diagnosis, anatomical site, histologic subtype, tumor grade, Crohn-like reaction and intratumoral lymphocytes. The immunohistochemical phenotype of microsatellite instability high colorectal cancer is poorly understood. Indeed, in bioptic material the 'traditional' multi-marker phenotype CK20/CK7/CDX2 is often applied to identify primary colorectal cancer or a colorectal cancer metastasis. However, its expression in colorectal cancer stratified by mismatch repair status has never previously been systematically analyzed.

In this study, a large series of colorectal cancers was stratified by immunohistochemistry into 1197 mismatch repair-proficient and 223 mismatch repair-deficient subgroups. Molecular testing was not performed on this cohort of patients as the sensitivity for determining microsatellite instability status by immunohistochemistry alone is reported to exceed 90% and the predictive value of absence of expression of either MLH1 or MSH2 for predicting microsatellite instability high has previously been described at 100%.⁴⁰ Additionally, microsatellite instability testing is recognized as infeasible in standard pathology laboratories and is currently not recommended in routine practice.⁴¹

One of the most obvious issues concerning the use of tissue microarrays is the extent of tumor heterogeneity, which may result from sampling of only one core of 0.6 mm in diameter. Bubendorf et al have reviewed this concern in depth.⁴² Several large studies in breast cancer, bladder carcinoma and with multiple tumor entities (multitumor array) have all reproduced findings from large whole tissue sections using tissue microarrays. They stress that the larger the number of tissues sampled, the less likely heterogeneity may play a role particularly as the likelihood of error in sampling will be similar in all tumor groups represented on one array. In addition, it should be underlined that tissue microarrays are not used for individual case assessment, but are designed for high throughput population-based phenotypic and genotypic evaluation.⁴³ Therefore, associations between molecular alterations and clinical and morphological parameters are likely to be similarly detectable on sufficiently large tissue microarrays as in large section analyses. Although it is possible that a small proportion of presumed mismatch repair-deficient colorectal cancer cases were incorrectly assigned, the overall findings are likely to be valid in view of the large numbers of samples and the good fit with clinico-pathological features.

Our results highlighted a significant difference in the mean percentage of CK20- and CDX2-positive cells in mismatch repair-proficient and -deficient colorectal cancer (P<0.001). Additionally, multi-marker combinations with CDX2 and CK20 negativity were more frequently associated with mismatch repair-deficient (19.3 and 21.6%) compared with mismatch repair-proficient colorectal cancer (6.7 and 7.5%). The two most frequent marker combinations other than CK20+/CK7- such as CK20-/CK7- and CK20+/CK7+ were observed in 16.8 and 7.6%, respectively.

In daily practice, the phenotype CK20-/CK7- is used to clarify the diagnosis of hepatocellular carcinoma, renal cell carcinoma, prostate carcinoma, squamous cell and small cell lung cancer as well as head and neck carcinoma. Our findings suggest that microsatellite instability high colorectal cancer should be added to this list of differential diagnoses expressing this combination of markers.

In addition, CK20+/CK7+ is frequently attributed to urothelial tumors, ovarian mucinous adenocarcinoma, pancreatic adenocarcinoma and cholangiocarcinoma.⁵ Again, microsatellite instability high colorectal cancer cannot be excluded from the panel of possible diagnoses exhibiting this immunohistochemical phenotype.

CDX2 negativity was observed in 21.6% of microsatellite instability high colorectal cancer and is therefore not likely to narrow the differential diagnosis, which can be made by CK20 and CK7. Although CDX2 is known to be expressed in adenocarcinomas of intestinal origin^{18, 44} and also in primary ovarian mucinous tumors,¹⁹ CDX2 expression has been described in tumors at various anatomic sites.^{19, 45, 46, 47, 48, 49, 50} On the basis of this ubiquity, negative CDX2 expression in adenocarcinoma cannot systematically rule out a diagnosis of colorectal cancer. Especially, the immunohistochemical analysis of a mucinous adenocarcinoma with negative CK20 expression in addition to negative CDX2 expression, located on the right side of the colon of a female patient must be underlined as a potential pitfall of misdiagnosis, where a metastasis could be diagnosed without consideration of a possible microsatellite instability high colorectal cancer.

Our findings are in agreement with the results of a previous study where 32% of the microsatellite instability high colorectal cancers were CK20 negative compared with 9% of mismatch repair-proficient colorectal cancer.³³ CK20 negativity in colorectal cancer has been reported to range between 0 and 9%.^{4, 11, 51, 52, 53, 54} Park et al¹¹ observed an association between CK20 loss, higher tumor grade and right-sided location and Hinoi et al⁵⁵ observed an association between loss of CDX2, microsatellite instability high status, right-sided tumor location and poor differentiation.

To study systematically the prognostic role of CK20, CK7 and CDX2 we additionally analyzed the associations of these markers with clinico-pathological findings including T stage, N stage, tumor grade, vascular invasion, tumor location and survival using ROC curve analysis. In mismatch repair-proficient colorectal cancer loss of CK20 was associated with higher tumor grade whereas loss of CDX2 showed an association with higher T stage, N stage, and tumor grade, with vascular invasion and right sidedness. In mismatch repair-deficient colorectal cancer there was an association between loss of CK20, higher tumor grade and right-sided location. CDX2 loss was only associated with right sidedness in this subgroup of tumors.

In both, mismatch repair-proficient and -deficient colorectal cancer subsets loss of CK20 was associated with higher tumor grade, whereas overexpression of this marker in mismatch repair-proficient colorectal cancer was found to be an independent adverse prognostic indicator. This apparent contradiction may be explained by several factors: first, higher tumor grade does not necessarily imply worse survival, as is underlined by the high percentage of microsatellite instability high colorectal cancer, which are poorly differentiated but have a relatively good prognosis compared with micro-satellite stable tumors; second, most cases of distant colorectal cancer metastases express CK20; third, we have shown in our study that loss of CK20 is associated with more abundant intratumoral lymphocytes, which could be linked to the observed more favorable prognosis; indeed CK20-negative cases with presence of intratumoral lymphocytes showed a better survival than CK20-positive colorectal cancers without intratumoral lymphocytes; fourth, using ROC curve analysis the cutoff for declaring a case CK20 'positive' was 85% in contrast to the often used cutoff of 0% (mainly for diagnostic purposes).

The focus of the present study was to elucidate the differential diagnostic and prognostic role of CK20, CK7 and CDX2 in colorectal cancer stratified by mismatch repair status. Loss of CDX2 and CK20 overexpression are markers of tumor progression in mismatch repair-proficient colorectal cancer. Loss of CDX2 and CK20 are more frequently encountered in mismatch repair-deficient colorectal cancer, which should be taken into consideration to correctly differentiate between primary and metastatic colorectal cancer.

References

1. Loy TS, Calaluce RD. Utility of cytokeratin immunostaining in separating pulmonary adenocarcinomas from colonic adenocarcinomas. Am J Clin Pathol 1994;102:764–767. | PubMed | ISI | ChemPort |
2. Rubin BP, Skarin AT, Pisick E, et al. Use of cytokeratins 7 and 20 in determining the origin of metastatic carcinoma of unknown primary, with special emphasis on lung cancer. Eur J Cancer Prev 2001;10:77–82. | Article | PubMed | ISI | ChemPort |
3. Tot T. The role of cytokeratins 20 and 7 and estrogen receptor analysis in separation of metastatic lobular carcinoma of the breast and metastatic signet ring cell carcinoma of the gastrointestinal tract. Apmis 2000;108:467–472. | Article | PubMed | ISI | ChemPort |
4. Kende AI, Carr NJ, Sobin LH, et al. Expression of cytokeratins 7 and 20 in carcinomas of the gastrointestinal tract. Histopathology. 2003;42:137–140. | Article | PubMed | ChemPort |
5. Varadhachary GR, Abbruzzese JL, Lenzi R, et al. Diagnostic strategies for unknown primary cancer. Cancer 2004;100:1776–1785. | Article | PubMed | ChemPort |
6. Osborn M, van Lessen G, Weber K, et al. Differential diagnosis of gastrointestinal carcinomas by using monoclonal antibodies specific for individual keratin polypeptides. Lab Invest 1986;55:497–504. | PubMed | ISI | ChemPort |
7. Gyure KA, Morrison AL. Cytokeratin 7 and 20 expression in choroid plexus tumors: utility in differentiating these neoplasms from metastatic carcinomas. Mod Pathol 2000;13:638–643. | Article | PubMed | ISI | ChemPort |
8. Lagendijk JH, Mullink H, Van Diest PJ, et al. Tracing the origin of adenocarcinomas with unknown primary using immunohistochemistry: differential diagnosis between colonic and ovarian carcinomas as primary sites. Hum Pathol 1998;29:491–497. | Article | PubMed | ISI | ChemPort |
9. Moll R, Franke WW, Schiller DL, et al. The catalog of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells. Cell 1987;31:11–24. | Article |
10. Moll R, Zimbelmann R, Goldschmidt MD, et al. The human gene encoding cytokeratin 20 and its expression during fetal development and in gastrointestinal carcinomas. Differentiation 1993;53:75–93. | Article | PubMed | ISI | ChemPort |
11. Park SY, Kim HS, Hong EK, et al. Expression of cytokeratins 7 and 20 in primary carcinomas of the stomach and colorectum and their value in the differential diagnosis of metastatic carcinomas to the ovary. Hum Pathol 2002;33:1078–1085. | Article | PubMed | ISI | ChemPort |
12. Roh MS, Hong SH. Utility of thyroid transcription factor-1 and cytokeratin 20 in identifying the origin of metastatic carcinomas of cervical lymph nodes. J Korean Med Sci 2002;17:512–517. | PubMed | ChemPort |
13. Moll R, Schiller DL, Franke WW, et al. Identification of protein IT of the intestinal cytoskeleton as a novel type I cytokeratin with unusual properties and expression patterns. J Cell Biol 1990;111:567–580. | Article | PubMed | ISI | ChemPort |
14. Moll R, Lowe A, Laufer J, et al. Cytokeratin 20 in human carcinomas. A new histodiagnostic marker detected by monoclonal antibodies. Am J Pathol 1992;140:427–447. | PubMed | ISI | ChemPort |
15. Silberstein E, Koretz M, Cagnano E, et al. Neuroendocrine (Merkel cell) carcinoma in regional lymph nodes without primary site. Isr Med Assoc J 2003;5:450–451. | PubMed |
16. Vang R, Gown AM, Barry TS, et al. Cytokeratins 7 and 20 in primary and secondary mucinous tumors of the ovary: analysis of coordinate immunohistochemical expression profiles and staining distribution in 179 cases. Am J Surg Pathol 2006;30:1130–1139. | PubMed |
17. Drummond F, Putt W, Fox M, et al. Cloning and chromosome assignment of the human CDX2 gene. Ann Hum Genet 1997;61:393–400. | Article | PubMed | ISI | ChemPort |
18. Werling RW, Yaziji H, Bacchi CE, et al. CDX2, a highly sensitive and specific marker of adenocarcinomas of intestinal origin: an immunohistochemical survey of 476 primary and metastatic carcinomas. Am J Surg Pathol 2003;27:303–310. | Article | PubMed | ISI |
19. Vang R, Gown AM, Wu LS, et al. Immunohistochemical expression of CDX2 in primary ovarian mucinous tumors and metastatic mucinous carcinomas involving the ovary: comparison with CK20 and correlation with coordinate expression of CK7. Mod Pathol 2006;19:1421–1428. | Article | PubMed | ChemPort |
20. Kinzler KW, Vogelstein B. Lessons from hereditary colorectal cancer. Cell 1996;87:159–170. | Article | PubMed | ISI | ChemPort |
21. Kambara T, Simms LA, Whitehall VL, et al. BRAF mutation is associated with DNA methylation in serrated polyps and cancers of the colorectum. Gut 2004;53:1137–1144. | Article | PubMed | ISI | ChemPort |
22. Kane MF, Loda M, Gaida GM, et al. Methylation of the hMLH1 promoter correlates with lack of expression of hMLH1 in sporadic colon tumors and mismatch repair-defective human tumor cell lines. Cancer Res 1997;57:808–811. | PubMed | ISI | ChemPort |
23. Jass JR. Classification of colorectal cancer based on correlation of clinical, morphological and molecular features. Histopathology 2007;50:113–130. | Article | PubMed | ChemPort |
24. Heinimann K, Muller H, Dobbie Z. Microsatellite instability in colorectal cancer. N Engl J Med 2000;342:1607–1608. | Article | PubMed | ChemPort |
25. Heinimann K, Muller H, Weber W, Scott RJ. Disease expression in Swiss hereditary non-polyposis colorectal cancer (HNPCC) kindreds. Int J Cancer 1997;74:281–285. | Article | PubMed | ChemPort |
26. Heinimann K, Scott RJ, Buerstedde JM, et al. Influence of selection criteria on mutation detection in patients with hereditary nonpolyposis colorectal cancer. Cancer 1999;85:2512–2518. | Article | PubMed | ChemPort |
27. Heinimann K, Scott RJ, Chappuis P, et al. N-acetyltransferase 2 influences cancer prevalence in hMLH1/hMSH2 mutation carriers. Cancer Res 1999;59:3038–3040. | PubMed | ChemPort |
28. Truninger K, Menigatti M, Luz J, et al. Immunohistochemical analysis reveals high frequency of PMS2 defects in colorectal cancer. Gastroenterology 2005;128:1160–1171. | Article | PubMed | ISI | ChemPort |
29. Westphalen AA, Russell AM, Buser M, et al. Evidence for genetic anticipation in hereditary non-polyposis colorectal cancer. Hum Genet 2005;116:461–465. | Article | PubMed | ChemPort |
30. Zhang J, Lindroos A, Ollila S, et al. Gene conversion is a frequent mechanism of inactivation of the wild-type allele in cancers from MLH1/MSH2 deletion carriers. Cancer Res 2006;66:659–664. | Article | PubMed | ISI | ChemPort |
31. Saunders M, Iveson T. Management of advanced colorectal cancer: state of the art. Br J Cancer 2006;95:131–138. | Article | PubMed | ISI | ChemPort |
32. Plasilova M, Zhang J, Okhowat R, et al. A de novo MLH1 germ line mutation in a 31-year-old colorectal cancer patient. Genes Chromosomes Cancer 2006;45:1106–1110. | Article | PubMed | ChemPort |
33. McGregor DK, Wu TT, Rashid A, Luthra R, Hamilton SR. Reduced expression of cytokeratin 20 in colorectal carcinomas with high levels of microsatellite instability. Am J Surg Pathol 2004;28:712–718. | Article | PubMed |
34. Sauter G, Simon R, Hillan K. Tissue microarrays in drug discovery. Nat Rev Drug Discov 2003;2:962–972. | Article | PubMed | ISI | ChemPort |
35. Hampel H, Stephens JA, Pukkala E, et al. Cancer risk in hereditary nonpolyposis colorectal cancer syndrome: later age of onset. Gastroenterology 2005;129:415–421. | Article | PubMed | ISI |
36. Zlobec I, Terracciano L, Jass JR, Lugli A. Value of staining intensity in the interpretation of immunohistochemistry for tumor markers in colorectal cancer. Virchows Arch 2007;45:763–769. | Article | ChemPort |
37. Zlobec I, Steele R, Terracciano L, Jass JR, Lugli A. Selecting immunohistochemical cut-off scores for novel biomarkers of progression and survival in colorectal cancer. J Clin Pathol 2007;60:1112–1116. | Article | PubMed |
38. Umar A, Boland CR, Terdiman JP, et al. Revised Bethesda Guidelines for hereditary nonpolyposis colorectal cancer (Lynch syndrome) and microsatellite instability. J Natl Cancer Inst 2004;96:261–268. | PubMed | ChemPort |
39. Jenkins MA, Hayashi S, O'Shea AM, et al. Pathology features in Bethesda guidelines predict colorectal cancer microsatellite instability: a population-based study. Gastroenterology 2007;133:48–56. | Article | PubMed | ChemPort |
40. Lindor NM, Burgart LJ, Leontovich O, et al. Immunohistochemistry versus microsatellite instability testing in phenotyping colorectal tumors. J Clin Oncol 2002;20:1043–1048. | Article | PubMed | ISI | ChemPort |
41. Jass JR. Re: Ward et al. Routine testing for mismatch repair deficiency in sporadic colorectal cancer is justified. J Pathol 2005;207:377–384. J Pathol 2006;208:590–1. | Article | PubMed |
42. Bubendorf L, Nocito A, Moch H, et al. Tissue microarray (TMA) technology: miniaturized pathology archives for high-throughput in situ studies. J Pathol 2001;195:72–79. | Article | PubMed | ISI | ChemPort |
43. Tzankov A, Went P, Zimpfer A, et al. Tissue microarray technology: principles, pitfalls and perspectives—lessons learned from hematological malignancies. Exp Gerontol 2005;40:737–744. | Article | PubMed | ChemPort |
44. Kaimaktchiev V, Terracciano L, Tornillo L, et al. The homeobox intestinal differentiation factor CDX2 is selectively expressed in gastrointestinal adenocarcinomas. Mod Pathol 2004;17:1392–1399. | Article | PubMed | ISI | ChemPort |
45. Mazziotta RM, Borczuk AC, Powell CA, et al. CDX2 immunostaining as a gastrointestinal marker: expression in lung carcinomas is a potential pitfall. Appl Immunohistochem Mol Morphol 2005;13:55–60. | Article | PubMed |
46. Ortiz-Rey JA, Alvarez C, San Miguel P, et al. Expression of CDX2, cytokeratins 7 and 20 in sinonasal intestinal-type adenocarcinoma. Appl Immunohistochem Mol Morphol 2005;13:142–146. | Article | PubMed | ChemPort |
47. Yatabe Y, Koga T, Mitsudomi T, et al. CK20 expression, CDX2 expression, K-ras mutation, and goblet cell morphology in a subset of lung adenocarcinomas. J Pathol 2004;203:645–652. | Article | PubMed | ChemPort |
48. Dennis JL, Hvidsten TR, Wit EC, et al. Markers of adenocarcinoma characteristic of the site of origin: development of a diagnostic algorithm. Clin Cancer Res 2005;11:3766–3772. | Article | PubMed | ChemPort |
49. Wu XS, Akiyama Y, Igari T, et al. Expression of homeodomain protein CDX2 in gallbladder carcinomas. J Cancer Res Clin Oncol 2005;131:271–278. | Article | PubMed | ChemPort |
50. Logani S, Oliva E, Arnell PM, et al. Use of novel immunohistochemical markers expressed in colonic adenocarcinoma to distinguish primary ovarian tumors from metastatic colorectal carcinoma. Mod Pathol 2005;18:19–25. | Article | PubMed | ChemPort |
51. Berezowski K, Stastny JF, Kornstein MJ, et al. Cytokeratins 7 and 20 and carcinoembryonic antigen in ovarian and colonic carcinoma. Mod Pathol 1996;9:426–429. | PubMed | ISI | ChemPort |
52. Kummar S, Fogarasi M, Canova A, et al. Cytokeratin 7 and 20 staining for the diagnosis of lung and colorectal adenocarcinoma. Br J Cancer 2002;86:1884–1887. | Article | PubMed | ChemPort |
53. Tot T. The value of cytokeratins 20 and 7 in discriminating metastatic adenocarcinomas from pleural mesotheliomas. Cancer 2001;92:2727–2732. | Article | PubMed | ISI | ChemPort |
54. Tot T. Cytokeratins 20 and 7 as biomarkers: usefulness in discriminating primary from metastatic adenocarcinoma. Eur J Cancer 2002;38:758–763. | Article | PubMed | ISI | ChemPort |
55. Hinoi T, Tani M, Lucas PC, et al. Loss of CDX2 expression and microsatellite instability are prominent features of large cell minimally differentiated carcinomas of the colon. Am J Pathol 2001;159:2239–2248. | PubMed | ISI | ChemPort |

Acknowledgements

This study was supported by a grant from the Novartis Foundation, formerly Ciba-Geigy-Jubilee-Foundation (I.Z.). We thank Privatdozent Dr Hanspeter Spichtin, Institute of Clinical Pathology Basel, Switzerland and Professor Dr Robert Maurer, Institute of Pathology, Stadtspital Triemli, Zuerich, Switzerland for providing the cases.