Cytokeratin 20 in Noncolorectal Lymph Nodes

Abstract & Commentary

Synopsis: Cytokeratin 20 (CK20) is a member of the intermediate filament protein family that has been studied as a specific marker for disseminated colorectal cancer cells. The specificity of CK20 was examined by a real-time PCR assay to determine CK20 expression in tumors and lymph nodes of subjects with colorectal and breast cancer, head and neck and vulvar squamous cell carcinoma, and melanoma. Low-level CK expression was detected in infiltrated lymph nodes from subjects with noncolorectal cancer types and suggest potential advantage of detecting circulating epithelial cells by quantitative PCR.

Source: Soong R, et al. Clin Cancer Res. 2001;7: 3423-3429.

Cytokeratin 20 (ck20) is a keratin protein with known distribution in epithelial cells of the gastrointestinal tract.1 Immunohistochemical studies have suggested potential use of CK20 as a means to separate adenocarcinomas of gastrointestinal vs. nongastrointestinal origin.1 However, more sensitive assays such as a CK20-reverse transcription (RT)-PCR assay, have reported positivity in several other types of cancer cells, including breast cancer cells.2 The CK20-RT-PCR assay has been used to detect disseminated colorectal cancer cells in lymph nodes, blood, and bone marrow and has been suggested to represent a means to significantly improve current staging protocols for colorectal cancer patients.3 The CK20-RT-PCR assay was used in combination with a CEA-RT-PCR assay to enhance ability to detect circulating tumor cells in a study involving 100 colorectal cancer patients (50 primary tumors only and 50 liver metastases) and 70 control individuals without known cancer.4 Circulating colorectal cancer cell positivity was detected in 48% of colorectal cancer patients with 1 sample of CEA alone, 34% with 1 sample of CK20 alone, and 74% when both assays and 3 samples were used to identify circulating cancer cells.4 However, 3 noncancer patients (4.3%) were positive for either CEA (2 patients) or CK20 (1 patient). It was suggested that RT-PCR-based studies for circulating cancer cells should involve multiple blood samples with identification of multiple tumor related cDNA products.4 Thus, additional study to determine the specificity and reproducibility of CK20 assays was needed.

This study by Soong and colleagues used 4 methods for RT-PCR quantification to clarify the expression profile and specificity of CK20 expression. Samples were obtained from 10 colorectal tumors and 11 breast tumors as well as 98 lymph nodes from various cancer types from subjects undergoing surgery for their disease. RT-PCR reactions were performed for CK20 and for the housekeeping gene, porphorynbilinogen deaminase (PBGD). Rigorous optimization experiments with the HT29 colon carcinoma call line and the HL60 promyelocytic leukemia cell line were performed to demonstrate sensitivity and specificity of these assays. CK20 could be detected in a dilution of 10 HT29 cells mixed with 1,000,000 HL60 cells, while no signal was detected from 1,000,000 HL60 cells alone. The reproducibility of each of the 4 methods for quantifying CK20 was assessed using a single tissue sample in a total of 27 replicates for each method, involving 3 serial dilutions, 3 separate reverse transcriptions, and 9 individual PCR assays. The coefficient of variation for each of the 4 methods was 43%, 32%, 12%, and 13%. Of the tumors evaluated, 10 of 10 (100%) of the colorectal cancers and 8 of 11 (73%) of the breast cancers were CK20 positive. Of the lymph nodes evaluated, 9 of 9 (100%) of the colorectal lymph nodes and 41 of 89 (46%) of the noncolorectal lymph nodes had detectable CK20 mRNA. A strong correlation existed between CK20 detection and lymph node epithelial invasion in the noncolorectal lymph nodes. The CK20 quantity, as assessed by the mean relative CK20 quantity, was significantly higher in colorectal than noncolorectal samples. Soong et al concluded that a standardizable, quantitative CK20 PCR assay can be used to characterize epithelial-specific, low-level CK20 expression in cancer types not expected to express CK20. The clinical implications of these findings require further analysis.

Comment by Mark R. Albertini, MD

The use of monitoring tools, such as detection of CK20 in blood, lymph node, or tumor samples, represent potential surrogate markers for biological activity of several cancers. Soong et al describe several essential issues that need to be addressed when evaluating a potential surrogate marker for cancer. The assay needs to be quantitative and needs to be reproducible. The assay should ideally use a small amount of clinical material and have some correlation with biological activity of the cancer. The assay should be reflective of the in vivo situation and require minimal in vitro manipulations. The assay should have some reference standard for purposes of comparison. The current report’s description of the characterization and evaluation of sensitivity and specificity of the CK20 RT-PCR assays fulfills many of these criteria. The evaluation of a biological assay requires some understanding of the assay accuracy, precision, specificity, and sensitivity. This careful description of assay standardization should be required of all reports describing potential surrogate markers.

The current report clearly identifies low-level, epithelial cell-specific CK20 expression in infiltrated lymph nodes from subjects with noncolorectal cancers. The discrepancies in prior reports likely resulted from variability in the sensitivity of the assays used for CK20 detection. While the level of CK20 expression appears to be much greater in colorectal samples, low-level expression was also present in noncolorectal samples. Thus, a reproducible assay exists and the biological relevance of CK expression in tumors, lymph nodes, and blood can now be addressed for several tumor types in larger prospective clinical studies. A careful description of CK20 assay characteristics will be an essential requirement for meaningful testing of this potential surrogate marker in clinical studies.


1. Miettinen M. Mod Pathol. 1995;8(4):384-388.

2. Bostick PJ, et al. J Clin Oncol. 1998;16(8):2632-2640.

3. Weitz J, et al. Clin Cancer Res. 1999;5(7):1830-1836.

4. Wharton RQ, et al. Clin Cancer Res. 1999;5(12): 4158-4163.

Dr. Albertini is Associate Professor, Department of Medicine, University of Wisconsin Medical School, Madison, WI.