Cytokeratin 5 and cytokeratin 6 expressions are unconnected in normal and cancerous tissues and have separate diagnostic implications

Cytokeratins (CKs) 5 and 6 are functionally unrelated but often analyzed together using bispecific antibodies in diagnostic immunohistochemistry. To better understand the diagnostic utility of CK5 or CK6 alone, tissue microarrays with > 15,000 samples from 120 different tumor types as well as 608 samples of 76 different normal tissues were analyzed by immunohistochemistry. In normal tissues, both CKs occurred in the squamous epithelium; CK5 dominated in basal and CK6 in suprabasal layers. CK5 (not CK6) stained basal cells in various other organs. Within tumors, both CK5 and CK6 were seen in > 95% of squamous cell carcinomas, but other tumor entities showed different results: CK5 predominated in urothelial carcinoma and mesothelioma, but CK6 in adenocarcinomas. Joint analysis of both CK5 and CK6 obscured the discrimination of epithelioid mesothelioma (100% positive for CK5 alone and for CK5/6) from adenocarcinoma of the lung (12.8% positive for CK5 alone; 23.7% positive for CK5/6). CK5 and CK6 expressions were both linked to high grade, estrogen receptor, and progesterone receptor negativity in breast cancer (p < 0.0001 each), grade/stage progression in urothelial cancer (p < 0.0001), and RAS mutations in colorectal cancer (p < 0.01). Useful diagnostic properties which are commonly attributed to CK5/6 antibodies such as basal cell staining in the prostate, distinction of adenocarcinoma of the lung from squamous cell carcinoma and epithelioid mesothelioma, and identification of basal-type features in urothelial cancer are solely driven by CK5. At least for the purpose of distinguishing thoracic tumors, monospecific CK5 antibodies may be better suited than bispecific CK5/6 antibodies. Supplementary Information The online version contains supplementary material available at 10.1007/s00428-021-03204-4.

To better understand the clinical utility of immunohistochemical analysis of CK5 and CK6 alone, both proteins were analyzed in more than 14,000 tumor tissue samples from 120 different tumor types and subtypes as well as 76 non-neoplastic tissue categories by immunohistochemistry (IHC) in a tissue microarray (TMA) format in this study.

Tissue microarrays (TMAs)
Our normal tissue TMA was composed of 8 samples from 8 different donors for each of 76 different normal tissue types (608 samples on one slide). The cancer TMAs contained a total of 15,966 primary tumors from 120 tumor types and subtypes. Histopathological data on pathological tumor stage (pT), histological grade, and pathological lymph node status (pN) were available from up to 2,075 breast, 1,663 bladder, 327 gastric, 598 pancreatic, 2,351 colorectal, 524 ovarian, and 259 endometrial cancers. Clinical follow-up data were available from 1,183 breast cancer and 254 urinary bladder cancer patients with a median follow-up time of 49/14 months (range 1-88/1-77). Molecular data on HER2, estrogen receptor (ER), and progesterone receptor (PR) status, microsatellite instability (MSI), and RAS mutations were available from previous studies [34,35]. The composition of both normal and cancer TMAs is described in the results section. All samples were from the archives of the Institutes of Pathology, University Hospital of Hamburg, Germany, the Institute of Pathology, Clinical Center Osnabrueck, Germany, and Department of Pathology, Academic Hospital Fuerth, Germany. Tissues were fixed in 4% buffered formalin and then embedded in paraffin. The TMA manufacturing process was described earlier in detail [36,37]. In brief, one tissue spot (diameter: 0.6 mm) was transmitted from a representative cancer-containing donor block in an empty recipient paraffin block. The use of archived remnants of diagnostic tissues for manufacturing of TMAs and their analysis for research purposes as well as patient data analysis were according to local laws (HmbKHG, §12) and approved by the local ethics committee (Ethics Commission Hamburg, WF-049/09). All work has been carried out in compliance with the Helsinki declaration.

Immunohistochemistry
Freshly cut TMA sections were immunostained on 1 day and in one experiment. Slides were deparaffinized with xylol, rehydrated through a graded alcohol series, and exposed to heat-induced antigen retrieval for 5 min in an autoclave at 121 °C in pH 9 DakoTarget Retrieval Solution™ (Agilent, CA, USA; #S2367,). Endogenous peroxidase activity was blocked with Dako Peroxidase Blocking Solution™ (Agilent, CA, USA; #52,023) for 10 min. Primary antibodies specific for CK5 protein (mouse monoclonal, clone MSVA-605 M) and CK6 protein (rabbit recombinant, clone MSVA-606R), both from MS Validated Antibodies, Hamburg, Germany were applied at a dilution of 1:150 each at 37 °C for 60 min. Bound antibody was then visualized using the EnVision Kit™ (Agilent, CA, USA; #K5007) according to the manufacturer's directions. For tumor tissues, the percentage of positive neoplastic cells was estimated, and the staining intensity was semiquantitatively recorded (0, 1 + , 2 + , 3 +). For statistical analyses, the staining results were categorized into four groups. Tumors without any staining were considered negative. Tumors with 1 + staining intensity in ≤ 70% of cells or 2 + intensity in ≤ 30% of cells were considered weakly positive. Tumors with 1 + staining intensity in > 70% of cells or 2 + intensity in 31-70% or 3 + intensity in ≤ 30% were considered moderately positive. Tumors with 2 + intensity in > 70% or 3 + intensity in > 30% of cells were considered strongly positive.

Statistics
Statistical calculations were performed with JMP 14 software (SAS Institute Inc., NC, USA). Contingency tables and the chi 2 -test were performed to search for associations between CK5 or CK6 and tumor phenotype. Survival curves were calculated according to Kaplan-Meier. The Log-Rank test was applied to detect significant differences between groups. A p-value of ≥ 0.05 was considered as statistically significant.

Technical issues
A total of 12,525 (78.5%) of 15,966 tumor samples were interpretable for CK5 and 12,898 (80.8%) of 15,966 tumor samples were interpretable for CK6 in this TMA analysis. The remaining samples were not analyzable due to the lack of unequivocal tumor cells or loss of the tissue spot during the technical procedures. In the normal tissue TMA, a sufficient number of samples was always interpretable per tissue type to determine CK5 and CK6 expressions.

CK5 in normal tissues
Strong CK5 immunostaining was seen in all keratinizing and non-keratinizing squamous epithelia (Fig. 1A) with a predominance of the staining in the basal cells of the epidermis, hair follicles, sebaceous glands, all epithelial cells of tonsil crypts, and of the thymus (Fig. 1B). In the urothelium, only the basal cell layers stained CK5 positive (Fig. 1C). CK5 immunostaining was also seen in myoepithelial cells and basal cells of excretion ducts of salivary and bronchial glands, basal cells of the prostate (Fig. 1D), seminal vesicle, respiratory epithelium, endocervix (not all glands), columnar cells (not all), and basal cells of the epididymis, myoepithelial cells of the breast. Amnion and chorion cells of the placenta also showed strong CK5 staining. CK5 immunostaining was absent in the lung, liver, pancreas, testis, kidney, gastrointestinal epithelial cells, Brunner glands, fallopian tube, adrenal gland, parathyroid gland, brain, adeno-and neurohypophysis, spleen, lymph node, all hematopoetic cell types, and all mesenchymal tissues.

CK6 in normal tissues
A preferential immunostaining in squamous epithelial cells was also seen for CK6 but the staining was most intense in suprabasal cell layers and basal cells were negative or only weakly stained (Fig. 1E). A strong CK6 positivity was also seen in hair follicles, sebaceous glands, a fraction of the squamous cells in tonsil crypts, corpuscles of Hassall's (but not in other epithelial cells) of the thymus (Fig. 1F), intercalated ducts of salivary and bronchial glands, basal cells and sometimes also ciliated cells (but not ciliae) of respiratory epithelium, few scattered endometrial cells (only in few samples), and in amnion/chorion cells (but not trophoblastic cells) of the placenta. CK6 immunostaining was absent in the urothelium (Fig. 1G), lung, liver, pancreas, prostate ( Fig. 1H), seminal vesicle, epididymis, testis, kidney, gastrointestinal epithelial cells, Brunner glands, fallopian tube, adrenal gland, thyroid, parathyroid gland, brain, adeno-and neurohypophysis, spleen, lymph node, all hematopoietic cell types, and all mesenchymal tissues.

CK5 and CK6 in tumors
Our tumor analysis revealed CK5 positivity in 21.1% (4.4% weak, 2.7% moderate, 14.0% strong) and CK6 positivity in  22.8% (8.4% weak, 4.1% moderate, 10.4% strong) of tumors. Of 2,920 CK5 and/or CK6 positive tumors, 1,921 (66%) showed staining for both CK5 and CK6, 461 (16%) stained positive only for CK5, and 538 (18%) showed positivity for CK6 alone. A detailed description of the results for CK5 and CK6 is given in Table 1. Representative images are shown in Fig. 2. Both antibodies showed positive staining in > 95% of squamous cell carcinomas from various origins. For most other tumor entities, CK5 and CK6 showed different positivity rates. CK5 was the predominant staining in mesothelioma, basal cell carcinoma of the skin, urothelial carcinoma, thymoma, and salivary gland tumors, while CK6 predominated in various adenocarcinomas. It was noticeable for all tumor entities that either CK5 or CK6 was the predominant staining and that adding the second cytokeratin did not relevantly increase the fraction of positive cases (Supplementary Fig. 1). A separate analysis of thoracic tumors (mesothelioma vs. squamous cell carcinoma of the lung vs. adenocarcinoma of the lung) revealed that the combined use of both CK5 and CK6 immunostaining hindered the discrimination of these tumors because the positivity rate of lung adenocarcinomas increased from 12.8% (CK5 alone) to 23.7% (CK5 and/or CK6 positive) while both squamous cell carcinomas and epithelioid mesotheliomas were positive in 100% irrespective of whether CK5 alone or CK5 and CK6 were jointly applied (Fig. 3).
A comparison with histopathological features of cancer aggressiveness and/or clinical follow-up data in 120 different tumor entities revealed only few statistical associations ( Table 2). Both CK5 and CK6 expressions were linked to high grade, estrogen and progesterone receptor negativity, and "triple negativity" in breast cancer (p < 0.0001 each), grade/stage progression in urothelial cancer (p < 0.0001), and RAS mutations in colorectal cancer (p < 0.01). CK5 expression was also associated with nodal metastasis in urothelial cancer (p = 0.0007) while CK6 expression was linked to nodal metastasis in gastric and ovarian cancer (p < 0.05). The CK5 and CK6 expression status was unrelated to overall patient survival in breast cancer (Supplementary Fig. 2a-c) and recurrence-free survival in patients treated by cystectomy for their urothelial carcinomas (Supplementary Fig. 2e-g).

Discussion
The successful analysis of a broad range of normal tissues and of more than 10,000 cancers for CK5 and CK6 by immunohistochemistry demonstrates important differences in the expression patterns of these cytokeratins. The data collected in this study suggest a superior diagnostic utility of monospecific CK5 or CK6 antibodies for immunohistochemical analysis as compared to bispecific CK5/6 antibodies. Additionally, literature data from various tumor types on CK5 expression ( Supplementary Fig. 3) or CK6 expression ( Supplementary Fig. 4) clearly demonstrate that such information cannot be easily obtained from the literature due to highly discrepant data across many studies.
That cytokeratins 5 and 6 have very different expression patterns is particularly demonstrated by the results of our extensive normal tissue analysis. Although both proteins were found in the majority of squamous epithelia, their staining patterns differed considerably with CK5 preferentially staining basal cells and CK6 preferably occurring in suprabasal layers. Given the complementary staining patterns of CK5 (basal) and CK6 (suprabasal), one might expect that the combined use of both antibodies could improve the positivity rate in squamous cell carcinomas of different sites of origin. That the combined analysis of CK5 and CK6 only increased the fraction of positive squamous cell carcinomas by 0-1% (average 0.1%) if compared to CK5 analysis alone and by 0-2% (average 0.5%) if compared to CK6 analysis alone does not provide strong evidence for superiority of using a combined CK5/6 antibody for the identification of squamous cell carcinomas, however.
The analysis of 109 non-squamous cell cancer tumor entities also did not suggest a particular rationale for combining anti-CK5 and anti-CK6 antibodies. The separate analysis of these antibodies showed for the vast majority of analyzed cancers that their CK5/6 positivity rate was largely driven by either CK5 or CK6 and that the addition of the other cytokeratin only minimally increased the positivity rate. Moreover, the positivity rate of both CK5 and CK6 was generally so low that the combination of both cytokeratins did still not result in a diagnostically useful information. Whether a tumor entity such as endometrioid carcinoma of the ovary is CK5 positive in 43% or CK5/6 positive in 49% does not impact the diagnostic information obtained by analyzing CK5 and/or CK6.
Especially in the case of thoracic tumors, the isolated CK5 analysis appears to be advantageous as compared to CK5/6 immunohistochemistry. For example, CK5 alone was positive in 100% of epithelioid mesotheliomas but in only 12.8% of lung adenocarcinoma and can thus be used in panels designed to distinguish these entities. Adding CK6 increases the positivity rate in lung adenocarcinoma to 23.7% and thus reduces the diagnostic potential for distinguishing mesothelioma from adenocarcinomas. CK6-positive adenocarcinomas also limit the utility of CK5/6 in the distinction of squamous cell carcinoma vs. adenocarcinoma of the lung which may be difficult and often requires the use of IHC panels. In another major application of CK5/6 IHC-the detection of basal cells in the prostate-CK5 is solely responsible for the beneficial effects, while CK6 is not staining any basal cells.  The availability of clinical follow-up data or histopathological data related to cancer aggressiveness enabled us to investigate the potential role of aberrant CK5 and CK6 expressions in 8 different cancer entities. The findings in bladder cancer further challenged the utility of combined CK5/6 analysis. In normal urothelium, CK5 is expressed in the basal cell layers. Because this basal layer staining is retained in the vast majority of non-invasive urothelial carcinomas, especially if they are of low grade, most pTa grade 1/2 tumors were scored as "weak" or "moderately" CK5 positive in our scoring system. In pTa grade 3 and especially in muscle-invasive urothelial carcinomas, the tumors often either completely lose CK5 expression or show CK5 expression in all cells. As a result, the fraction of CK5 negative and of strongly positive cases increased markedly with bladder cancer grade and stage. A link between CK5 expression and bladder cancer progression was earlier reported by several authors [38][39][40][41]. Most of these studies have employed antibodies against CK5/6 and it has been assumed that CK5/6 positivity reflects a "basal-type" molecular subgroup of urothelial carcinoma [42,43]. Our separate analysis of CK6 revealed, however, that CK6 is unrelated to urothelial basal cells. CK6 upregulation has been considered a feature of squamous cell differentiation in urothelial cancer by others [44,45].
That both CK5 and CK6 expressions were statistically linked to high grade in breast cancer is consistent with data from various earlier studies [46][47][48]. Accordingly, CK5/6 expression is an established feature of basal-type breast cancer which is well known for its poor clinical outcome [49]. These findings may reflect a general phenomenon. Multiple studies have described a tendency towards a poor prognosis and/or unfavorable tumor phenotype in cancers that show de novo expression of a cytokeratin which is not expressed in its normal cell type of origin [50][51][52][53][54][55]. An altered expression pattern     of intermediate filaments appears to represent a common feature of cancer cell dedifferentiation that occurs during cancer progression and will thus be linked to unfavorable tumor features [56]. These significant associations with relevant histological tumor aspects were not found in endometrial, ovarian, stomach, pancreatic, and colon cancers, and argue against a major role of CK5 and CK6 expressions for tumor progression. The relationship between CK5 and CK6 expression and KRAS remains unclear as direct or indirect interactions between these proteins are not known and also not expected based on the individual functions of these two proteins.
In summary, our data show that important properties which are commonly attributed to CK5/6 antibodies such as basal cell staining in the prostate, distinction of mesothelioma and squamous cell carcinoma from adenocarcinoma of the lung, and identification of basal-type features in urothelial cancer are solely driven by CK5. At least for the purpose of distinguishing thoracic tumors, monospecific CK5 antibodies may be better suited than bispecific CK5/6 antibodies in diagnostic immunohistochemistry.