Keratin 7 expression in lymph node metastases but not in the primary tumour correlates with distant metastases and poor prognosis in colon carcinoma

Introduction

Colorectal carcinoma (CRC) is the third most common malignancy in humans and the fourth cause of cancer-related deaths [1]. Patients’ overall survival is dependent on several factors, e.g. tumour grade and stage, including lymph node involvement [2, 3]. Colorectal carcinomas with lymph node metastases (LNM) differ in many aspects from those that do not present nodal dissemination. The former show worse prognosis, a higher recurrence rate, and shorter overall and disease-free survival [4, 5]. Additionally, application of adjuvant systemic therapy also depends on the lymph node involvement [6].
The tumour cell characteristics of LNMs differ in many aspects from those in the primary tumours (PT) in many human malignancies, e.g., epithelial to mesenchymal transition (EMT) markers in breast carcinoma were more strongly expressed in metastatic lesions [7]. In addition, EMT phenotype more strongly correlated with prognosis if the LNM profile rather than that of the primary tumour was taken into account [8]. In 53% of colorectal carcinomas the histological grade differs between primary tumours and their respective LNMs [4]. However, the other cellular characteristics between the primary tumour and secondary deposits in CRC have not been studied extensively.
Keratins (Ks) form intermediate filaments that are characteristic for epithelial cells. Their family is composed of more than 80 types of proteins, which form hetero-dimers, composed of one molecule of type I K and another one of type II K [9]. Normal epithelial cells usually have a specific keratin profile that differs among various histological types of epithelium. Neoplasms largely retain the keratin profile during malignant transformation, and this feature is widely used in the histological differential diagnosis of carcinomas. Aberrant expression of Ks confers a poor prognosis for the patients, e.g., K8/18 in oral squamous cell carcinoma [10] and oesophageal carcinoma [11], K7 in oesophageal carcinoma [12] and K19 in hepatocellular carcinoma [13]. K20 is constantly expressed in the colon epithelium and thus it is a marker of tissue derivation in CRC. In contrast, K7 is not expressed in the normal colonic epithelium and its expression is an exceptional finding in CRC. However, some clinic-pathological subtypes, e.g., BRAF-mutated microsatellite stable or ulcerative-colitis associated CRCs, show much more frequent K7-positivity [14, 15].
In CRC, K7 expression in PT was substantially more frequent in tumours that disseminated to the LN (25.3%) compared to cases without LN involvement (17.3%) [16]. In addition, K7-positivity was associated with shorter survival [17]. However, to the best of our knowledge, no studies comparing K7 expression and its prognostic impact in PT versus their LNM in CRC have been published so far.
The aim of our study was to evaluate the prognostic significance of K7 in two independent cohorts of patients with CRC: one with and another without LN involvement. Additionally, we decided to compare the expression of K7 in PT and LNM in CRC to verify their prognostic value.

Material and methods

All cases were retrieved from the files of the Department of Pathomorphology, Medical University of Gdansk, Poland and the Centre of Oncology, Bydgoszcz, Poland These patients were operated on the Surgical Clinics in the years 1998-2004. The first cohort consisted of 129 patients with CRC disseminated to the regional lymph nodes (LN+). It included 54 females (42%) and 75 males (58%), with the mean age of 63.4 years (range 32-91). The mean follow-up period was 36.6 months (range 19-148.6).
The second cohort of LN-negative CRC (LN–) consisted of 85 patients (50 males and 35 females) with mean age of 66 years (range 32-87) and mean follow-up of 59 months (range 19-143). In 40 patients of this 85 LN– group, the tumour relapsed as a metastatic disease (lungs, liver, bones and skin) during the first three years following surgery. None of the patients received neoadjuvant chemotherapy. We decided to exclude rectal carcinoma from our analysis. Therefore, the cases of resected tumours from the large intestine only (sigmoid, traverse, right-side colon, left-side colon) were taken into account.
Basic clinic-pathological characteristics of both groups are shown in Table I.

Tissue microarrays

Tissue microarrays (TMAs) were constructed from the archival formalin-fixed, paraffin-embedded tissue blocks using a manual tissue arrayer (Beecher Instruments, MTAI, K7 BioSystems).
In the LN+ group, 3 areas rich in tumour cells from each case were punched from the PT and metastatic LN and transferred to the recipient block. In the case of multiple metastatic lymph nodes cores were taken haphazardly from up to three involved lymph nodes In the LN– group two areas of PT were taken from each case.

Immunohistochemistry

Immunohistochemical staining was performed on a DAKO Autostainer with antibody against K7 (DAKO, Clone OV-TL 12/30, ready-to-use).
Intensity of the immunohistochemical reaction was evaluated using the h-score system. In h-score, the respective intensity grades of reaction and the percentage of positive cells were multiplied and summed up (1x n%+2xn%+3xn% = y). Therefore, theoretically the score ranged from 0 to 300. The tumour tissue from all cores of a single case was evaluated, separately for PT and LNM, if present. The cytoplasmic reaction was regarded as positive K7 expression. For statistical purposes all cases with any level of K7 expression were referred to as positive. Each case was evaluated independently by two pathologists (P.C. and A.G.).

Statistics

Statistica software (version 12, StatSoft) was used for the analysis. 2 or Fisher’s exact test was applied for categorical data, where appropriate. The Kaplan-Meier estimator was employed for survival analysis, and the generated curves were compared with Cox’s F-test. The endpoint for the study was overall survival (OS). OS was defined as the time from sample collection to death or censoring. Censoring was defined as loss of follow-up or alive at the end of follow-up. Statistical significance was assumed when p ≤ 0.05. , being a measure of the strength of agreement, was calculated using MedCalc software (version 12.5.0.0).

Results

Keratin 7 expression is infrequent in colorectal carcinoma and shows low concordance between PT and LN metastases

In the LN+ group seven PTs (5.4%) and seven LNMs displayed K7 expression in the PT and in the lymph node metastases. In three cases K7 positivity was identified both in PT and LN, and in four cases expression was found either in PT or LN. Overall concordance between PT and LN was 94% ( = 0.396) (Table II).
Representative examples of K7 expression in PT and LNM are shown in Figures 1A,B and 2A,B.

Prognostic significance of keratin 7 expression

Keratin 7 expression in LNM correlated with shorter OS (p = 0.047) (Fig. 3). However, the survival was not altered in cases with K7 expression in PT (p = 0.3) (Fig. 4).
Distant metastases were more frequent in patients showing K7-positive LNM (4/23 vs. 3/104, p = 0.005), whereas the status of K7 in PT was not associated with metastatic dissemination (p = 0.44).
Keratin 7 expression was not different between analysed cohorts (LN+, LN–). K7 was expressed in eight cases (9.4%) in the LN– group, and this feature was associated neither with recurrence of disease (p = 0.26) nor with the overall survival of the patients (p = 0.23). (supplementary material).

Keratin 7 and clinic-pathological factors

Among the features analysed K7 expression was more frequent in grade 3 tumours in the LN– (p = 0.046) cohort. In the LN+ cohort K7 expression in LNM was associated with grade 3 (p = 0.049), while there was not such a correlation for PT (p = 0.29). Keratin 7 expression was not associated with sex, tumour stage, LN status in the LN+ cohort or histological type.

Discussion

In our study K7 expression in CRC was uncommon regardless of the LN involvement by cancer (LN+ vs. LN–, 5.4% and 9.41%, respectively). This is generally in line with the findings of the large study by Harbaum et al., who found K7 expression in 9% of unselected PT in a group of 370 patients [17]. However, the percentage of K7+ tumours can be much higher in certain molecular subtypes of CRC, e.g., K7 is expressed in 39% of cases in BRAF-mutated microsatellite-stable CRCs [14]. Similarly, pathogenic background may also determine this feature, since K7 positivity was identified in up to 59% of ulcerative colitis-associated CRC [15]. Seemingly, our cohorts of unselected CRC did not have overrepresentation of these clinic-molecular subgroups.
Interestingly, in a study of Bayrak et al., K7 expression in PT was more frequent in LN+ (25.3%) than in LN– (11%) CRC patients [16]. We did not observe such differences in K7 expression between these cohorts. So far, data about expression of K7 in LNM of CRC have not been reported. We are the first to show that K7 positivity in LNM occurs infrequently, with similar incidence to that present in PT (5.4%). Aberrant expression of keratins is a poor prognostic factor in multiple tumour types [10, 13], e.g., K7 correlates with an unfavourable clinical course in oesophageal carcinoma [12]. However, data about the poor prognostic role of K7 are limited. The only study that shows worse survival in CRC is the analysis of a large cohort performed by Harbaum et al. [17]. They found that 48% of patients with K7-positive tumours died of the disease compared to 33% with K7-negative tumours (p = 0.06). Despite the large size of the cohort (370 cases), the diffrenece was not statistically significant. This supports to some extent our observations indicating that expression of K7 in PT tumour is not a prognostic factor in CRC. However, we found that a potential poor prognostic factor is expression of CK7 in local nodal metastases of CRC.
In general, there is a high level of discordance in the expression of markers between PT and LNM in CRC. This concerns p53 and c-myc [18], microsatellite instability (MSI), CpG island methylator phenotype (CIMP) [19], p21, cyclin D1 [20] and thymidylate synthase [21]. IHC expression of tyrosine phosphatase type IV A member 3 (PTP4A3) was found in 18.4% of primary tumours and 91.6% of LN metastases. Mutations in p53 were also more frequent in LNM than in PT [22]. In LNM of CRC expression of markers responsible for more aggressive clinical behaviour is frequently stronger than in PT. This is the case with p53 [18, 23] and the Ki-67 index [23].
Expression of many markers in LNM, including thymidylate synthase [24], epidermal growth factor receptor (EGFR) [25] and p16 [26], shows better prognostic information in LNM than in PT of CRC. Similar observations were found in other cancers, for example in breast carcinoma [8].
In our LN+ group we observed a strong correlation between K7 positivity in LNM and the presence of distant metastases at the time of diagnosis (p = 0.006). This observation also confirms higher biological aggressiveness of tumours expressing K7 in LNM. Indeed, poor prognosis in these patients may be secondary to higher clinical stage of these tumours; however, multivariate analysis did not give support to such a conclusion. On the other hand, an analysis of the population of tumour cells that had disseminated to secondary sites provides additional information on the more aggressive component of the neoplasm. Identification of a potential prognostic marker in this selected clone may provide more precise data about the clinical course of the disease.
As K7 expression in LNM correlates with presence of distant metastases it would be interesting to investigate the expression of K7 in distant metastases. Unfortunately, this material is not readily available. Further studies comparing expression of potential markers, including K7 in PT, LNM and distant metastases of CRC could shed more light on the mutual relationships of these proteins in different tumour locations.

Conclusions

To summarise, we have described K7 expression in LNM but not in PT as a poor prognostic factor in CRC. K7 expression in LNM but not in PT correlates with presence of distant metastases in CRC. Low concordance in the expression of K7 between PT and LNM and its different prognostic influence confirms data on substantial changes in the biology of cancer cells in PT and LNM of CRC.

This study was supported by Grant No. ST-95 from the Medical University of Gdańsk.
PC and MB were supported by the programme “Mistrz” from the Foundation for Polish Science.
P.C., M.B. and M.S. were supported by Grant No. 2011/03/b/nz5/00519 (OPUS 2) from the National Centre of Science.
The authors declare no conflict of interest.

References

1. Fitzmaurice C, Dicker D, Pain A, et al. The Global Burden of Cancer 2013; JAMA Oncol 2015; 1: 505-527.
2. Lin HH, Yang HL, Lin JK, et al. The Number of risk factors determines the outcome of stage II colorectal cancer patients. Hepatogastroenterology 2014; 61: 1024-1027.
3. Parnaby CN, Scott NW, Ramsay G, et al. Prognostic value of lymph node ratio and extramural vascular invasion on survival for patients undergoing curative colon cancer resection. Br J Cancer 2015; 113: 212-219.
4. Resch A, Harbaum L, Pollheimer MJ, et al. Grading lymph node metastasis: a feasible approach for prognostication of patients with stage III colorectal cancer. J Clin Pathol 2015; 68: 742-745.
5. Resch A, Langner C. Lymph node staging in colorectal cancer: old controversies and recent advances. World J Gastroenterol 2013; 19: 8515-8526.
6. Benson AB, Bekaii-Saab T, Chan E, et al. Metastatic colon cancer, version 3.2013: featured updates to the NCCN Guidelines. J Natl Compr Canc Netw 2013; 11: 141-152.
7. Markiewicz A, Ksiazkiewicz M, Seroczynska B, et al. Heterogeneity of mesenchymal markers expression-molecular profiles of cancer cells disseminated by lymphatic and hematogenous routes in breast cancer. Cancers (Basel) 2013; 5: 1485-1503.
8. Markiewicz A, Ahrends T, Welnicka-Jaskiewicz M, et al. Expression of epithelial to mesenchymal transition-related markers in lymph node metastases as a surrogate for primary tumor metastatic potential in breast cancer. J Transl Med 2012; 10: 226.
9. Karantza V. Keratins in health and cancer: more than mere epithelial cell markers. Oncogene 2011; 30: 127-138.
10. Fillies T, Werkmeister R, Packeisen J, et al. Cytokeratin 8/18 expression indicates a poor prognosis in squamous cell carcinomas of the oral cavity. BMC Cancer 2006; 6: 10.
11. Makino T, Yamasaki M, Takeno A, et al. Cytokeratins 18 and 8 are poor prognostic markers in patients with squamous cell carcinoma of the oesophagus. Br J Cancer 2009; 101: 1298-1306.
12. Oue N, Noguchi T, Anami K, et al. Cytokeratin 7 is a predictive marker for survival in patients with esophageal squamous cell carcinoma. Ann Surg Oncol 2012; 19: 1902-1910.
13. Govaere O, Komuta M, Berkers J, et al. Keratin 19: a key role player in the invasion of human hepatocellular carcinomas. Gut 2014; 63: 674-685.
14. Landau MS, Kuan SF, Chiosea S, Pai RK. BRAF-mutated microsatellite stable colorectal carcinoma: an aggressive adenocarcinoma with reduced CDX2 and increased cytokeratin 7 immunohistochemical expression. Hum Pathol 2014; 45: 1704-1712.
15. Tatsumi N, Kushima R, Vieth M, et al. Cytokeratin 7/20 and mucin core protein expression in ulcerative colitis-associated colorectal neoplasms. Virchows Arch 2006; 448: 756-762.
16. Bayrak R, Yenidunya S, Haltas H. Cytokeratin 7 and cytokeratin 20 expression in colorectal adenocarcinomas. Pathol Res Pract 2011; 207: 156-160.
17. Harbaum L, Pollheimer MJ, Kornprat P, et al. Keratin 7 expression in colorectal cancer – freak of nature or significant finding? Histopathology 2011; 59: 225-234.
18. Zalata KR, Elshal MF, Foda AA, Shoma A. Genetic dissimilarity between primary colorectal carcinomas and their lymph node metastases: ploidy, p53, bcl-2, and c-myc expression – a pilot study. Tumour Biol 2015; 36: 6579-6584.
19. Messick CA, Church JM, Liu X, et al. Stage III colorectal cancer: molecular disparity between primary cancers and lymph node metastases. Ann Surg Oncol 2010; 17: 425-431.
20. McKay JA, Douglas JJ, Ross VG, et al. Cyclin D1 protein expression and gene polymorphism in colorectal cancer. Aberdeen Colorectal Initiative. Int J Cancer 2000; 88: 77-81.
21. Marsh S, McKay JA, Curran S, et al. Primary colorectal tumour is not an accurate predictor of thymidylate synthase in lymph node metastasis. Oncol Rep 2002; 9: 231-234.
22. Zhang JS, Caplin S, Bosman FT, Benhattar J. Genetic diversity at the p53 locus between primary human colorectal adenocarcinomas and their lymph-node metastases. Int J Cancer 1997; 70: 674-678.
23. Meteoglu I, Erdogdu IH, Tuncyurek P, et al. Nuclear factor kappa B, matrix metalloproteinase-1, p53, and Ki-67 expressions in the primary tumors and the lymph node metastases of colorectal cancer cases. Gastroenterol Res Pract 2015; 2015: 945392.
24. Ohrling K, Edler D, Hallstrom M, et al. Detection of thymidylate synthase expression in lymph node metastases of colorectal cancer can improve the prognostic information. J Clin Oncol 2005; 23: 5628-5634.
25. Deng Y, Kurland BF, Wang J, et al. High epidermal growth factor receptor expression in metastatic colorectal cancer lymph nodes may be more prognostic of poor survival than in primary tumor. Am J Clin Oncol 2009; 32: 245-252.
26. Karamitopoulou E, Zlobec I, Koumarianou A, et al. Expression of p16 in lymph node metastases of adjuvantly treated stage III colorectal cancer patients identifies poor prognostic subgroups: a retrospective analysis of biomarkers in matched primary tumor and lymph node metastases. Cancer 2010; 116: 4474-4486

Address for correspondence

Piotr Czapiewski
Department of Pathomorphology
Medical University of Gdańsk
Debinki 7
80-952 Gdańsk, Poland
e-mail: czapiewskipiotr@gumed.edu.pl