Claudin-1 Protein Expression is a Prognostic Marker of Patient Survival in Renal Cell Carcinomas

Claudins are integral cell membrane proteins physiologically found in tight junctions. The claudin family comprises 24 members, and many of these are dysregulated in human malignancies. Physiologically, claudins contribute to the function of tight junctions as a selective permeable barrier essentially important in epithelial tissues of the gastrointestinal tract, liver, and kidney. Tight junctions mediate physical adhesion and paracellular transport and also regulate signaling pathways in cellular proliferation and differentiation (1–4). Modified expression of claudins on mRNA and protein level has been implicated in the pathogenesis of several human diseases, including edema, diarrhea, and tumor metastasis (5). Tight junction proteins interact with actin and microtubules in the cytoskeleton and recruit other signaling molecules, e.g., from the RAF-1 and TGF-β pathways (4, 6, 7). Particular focus in claudin research concentrates on renal diseases because segment-specific claudin compositions are thought to directly determine the physiologic functions (8). Further examples of claudin expression–related diseases are inflammatory bowel disease, retinopathies, macular degeneration, and hepatitis (9–11).

Alterations of claudin expression in cancer tissues have been identified for several tumor entities including cancer of the bladder, prostate, breast, colon, ovary, and the thyroid. In these tumors, claudin dysregulation has been found associated with invasiveness, metastatic behavior, tumor recurrence, and shortened overall survival (12–20). Possibly, aberrant expression patterns of claudins may lead to a loss of cell polarity, diminished cell adhesion, and thus an invasive phenotype.

Thus far, relatively little is known about claudin expression in renal tumors. Renal cell carcinoma is the most common solid lesion within the kidney and is among the most common and lethal cancers in the United States (21). Despite the increased detection rate of small tumors by improved imaging techniques the mortality has not relevantly decreased. Until now only few research groups have investigated the expression of other claudins in renal carcinoma at DNA or RNA level (22–24). Claudin-7 and claudin-8 are relatively overexpressed in chromophobe renal cell carcinoma and oncocytomas (25, 26). Among the numerous molecular markers being currently investigated for renal cell carcinoma, claudins emerged as promising targets for diagnosis and future anticancer drug therapy (27, 28).

Claudin-1 is a 23-kDa transmembranous protein widely expressed in liver and kidney tissue. Functional analysis found claudin-1 to be a major constitutive component and crucial for the tight junction function. By transfecting claudin-1 and claudin-2 into tight junction deficient cells, tight junctions could be restored (29). This further supports the paramount importance of these two proteins for tight junction formation. Renal tissue damage by contrast media resulted in a possibly compensatory up-regulation of claudin-1, and deficiency of claudin-1 resulted in epidermal water loss and neonatal death in mice (30, 31). Germline mutations of claudin-1 have been associated with neonatal sclerosing cholangitis (32). Thus, the presence of claudin-1 in tight epithelia underlines its role in reducing paracellular permeability (33).

For claudin-1, like for other claudins, up- and down-regulation have recently been described in different solid malignancies with different tumorbiological and prognostic associations (12, 18–20, 34–36).

In this study, we assessed the expression of claudin-1 protein in renal cell carcinomas and correlated these findings with clinicopathologic parameters including patient survival.

Patient and tissue selection. We constructed a tissue microarray from renal cell carcinomas diagnosed at the Institute of Pathology, Charité-Universitätsmedizin Berlin, Campus Mitte, from 1993 until 2004 who underwent surgery for renal cell carcinoma at the Department of Urology, Charité-Universitätsmedizin Berlin. Cases were selected according to tissue availability, without any further stratification for clinical or pathologic prognostic factors. Patients with accompanying secondary malignancies were excluded. Three hundred eighteen renal cell carcinoma cases were represented on the tissue microarray. Staging met the International Union against Cancer 2002 criteria. Histologic classification and grading were according to WHO (37). The median patient age of these patients was 61 y (range, 30-86 y). Of the patients, 213 were men, 105 women. The majority of the carcinomas (n = 278, 87.4%) were of clear cell type. The remaining renal cell carcinomas were of papillary (n = 30; 9.4%) and chromophobe (n = 10; 3.1%) types. The pT status of these cases were as follows: 181 pT₁ (56.9%), 28 pT₂ (8.8%), 106 pT₃ (33.3%), and 3 pT₄ (0.9%). The Fuhrman grades were G₁ in 39 (13.3%), G₂ in 229 (72.0%), G₃ in 48 (15.2%), and G₄ in 2 (0.6%) cases. A nodal status was evaluable in 190 cases (59.7%) with positive nodal status in 20 cases (6.3%). Patients with papillary renal cell carcinoma had a higher frequency of lymph node metastases than those with clear cell renal cell carcinoma (4 of 30 versus 15 of 278, respectively). Metastasis, pT status, Fuhrman grade, performance status, age, and disease-specific death rate were similar for both histologic tumor types.

Distant metastasis status was determined preoperatively (computed tomography of chest, abdomen, and pelvis) and revealed 26 patients with synchronous distant metastasis (M₁).

Follow-up data were available for all 318 patients with a median follow-up time of 99 mo (range, 0-177 mo). Disease-specific death occurred in 106 patients (33.3%) after a median survival time of 31 mo (range, 0-151 mo). The clinicopathologic parameters, including the Eastern Cooperative Oncology Group performance status, are shown in Table 1.

For further validation of our findings on claudin-1 expression in papillary renal cell carcinoma, we subsequently analyzed 44 papillary renal cell carcinomas diagnosed at the UniversitätsSpital Zurich from 1993 and 2003. Thirty patients were men (68.2%), 14 women were (31.8%), and the median age was 58 y. The pT status was pT₁ for 30 (68.2%), pT₂ for 7 (16.9%), and pT₃ for 7 (16.9%) patients. Fuhrman grades were G₂ for 23 (52.3%), G₃ for 17 (38.6%), and G₄ for 4 (9.1%) patients. Five patients (11.3%) had a positive nodal status and for 22 patients (50%) no nodal status was available. Data on Eastern Cooperative Oncology Group performance status were not available. Synchronous distant metastasis was documented for 6 patients (13.6%) and no information on distant metastasis was available for 23 patients (52.3%). The median follow-up time (available for 38 patients) was 34 mo (range, 0-140 mo). Disease-specific death occurred in 13 patients (35.1%) during follow-up.

Tissue microarray construction. A tissue microarray was constructed as described before (38). We used a commercially available tissue arrayer (Beecher Instruments) and applied a punch diameter of 1 mm. Each case was represented by four (2× carcinoma; 2× normal renal tissue) cores. The normal tissue was taken from the medullary and the cortex region of the kidney for every case, if feasible. The whole tissue microarray was accomplished on six paraffin blocks.

In the additional tissue microarray from Zurich each papillary renal cell carcinoma was represented by one tissue core (0.6 mm).

Immunohistochemistry. The tissue microarray was freshly cut (3 μm) and mounted on superfrost slides (Menzel Gläser). After deparaffinization with xylene and gradual rehydration, antigen retrieval was achieved by pressure cooking in 0.01mol/L citrate buffer for 5 min. Slides were incubated with the primary antibody (rabbit polyclonal anticlaudin-1; Invitrogen Zymed; dilution 1:50) at room temperature for 1 h.

As negative controls we used whole tissue slides omitting the primary antibody. Detection took place by the REAL EnVision detection system (Dako) with diaminobenzidine peroxidase serving as chromogen. The slides were briefly counterstained with hematoxylin and aqueously mounted.

Evaluation. The immunohistologic stainings were evaluated by two genitourinary pathologists, blinded for patient outcome, on a multiheaded microscope. Discrepant cases were discussed until consensus was reached. The staining intensity was evaluated with a 4-tier grading system (0, negative; 1, weak; 2, moderate; and 3, strong staining intensity). We used a 10% threshold for staining positivity. For further statistical analyses the claudin-1 positive tumors were grouped against the negative ones.

Statistical analysis. Statistical analysis was done using SPSS, version 15.0 (SPSS Inc.). Correlations were calculated according to Spearman's rank order correlation. Fisher's exact and χ² tests were applied to assess the statistical significance of the associations between claudin-1 expression and clinicopathologic parameters. Univariate survival analysis was carried out according to Kaplan-Meier. Differences in survival curves were assessed with the Log rank test. Multivariate analyses were calculated according to the Cox regression model. P values <0.05 were considered significant.

Immunohistochemistry. Claudin-1 was immunohistochemically detected in the majority of the normal renal tissue, especially in the Bowman membrane of the glomeruli and slightly heterogeneously in the epithelium of distal tubuli and in the collecting ducts (Fig. 1A and B). Stromal cells and glomeruli were consistently negative.

The majority (n = 223) of renal cell carcinomas were negative for claudin-1. Weak, moderate, and strong staining was observed in 64, 19, and 12 cases, respectively. In claudin-1 positive cases (n = 95; 29.9%) the signal was mainly located at the cell membrane (n = 88; 27.7%). In some tumors, however, especially of the papillary subtype, an additional cytoplasmic staining was observed (n = 30; 9.4%).

About one fourth (68 of 278) of the clear cell renal cell carcinomas were positive for claudin-1 (Fig. 1C and D), whereas in more than three quarters of the papillary renal cell carcinomas (23 of 30; 76.6%) claudin-1 expression could be shown (Fig. 1E and F; Table 1). The high rate of positivity in papillary renal cell carcinoma was validated in the supplementary tissue microarray where 38 of 44 (86.4%) papillary renal cell carcinomas were positive for claudin-1.

Statistical associations and survival analyses. If all histologic subtypes were included, claudin-1 expression was significantly associated with positive nodal status (Table 1). As expected, claudin-1 was highly significantly associated with the papillary renal cell carcinoma subtype (Table 1). A trend was detectable for claudin-1-positivity in more poorly differentiated and metastatic tumors but failed statistical significance if tested for the whole cohort (Table 1). Tumor size/extension (pT) and age were not related to claudin-1 expression but surprisingly gender was (P < 0.001). In a stratified analysis according to histologic subtypes (clear cell renal cell carcinoma versus papillary renal cell carcinoma), none of the mentioned parameters reached significance in the group of papillary renal cell carcinoma (n = 30; all P values > 0.200). Meanwhile, for the clear cell renal cell carcinoma group (n = 278) pT status and age remained unrelated to claudin-1 expression whereas nodal status lost its significance (P = 0.110). The positive association of male gender with claudin-1 expression remained highly significant. Interestingly, in clear cell renal cell carcinoma the presence of metastasis and higher Fuhrman grade became significantly associated with claudin-1 expression with P = 0.049 and P = 0.024, respectively. These results further support the notion that claudin-1–positive tumors are more aggressive and exhibit a poorer differentiation.

The respective statistical associations of the papillary renal cell carcinomas from Zurich are described in Table 2. In these cases we also detected a predominance of claudin-1 expression in men compared with women. Additionally, in papillary renal cell carcinoma loss of claudin-1 was significantly associated with parameters of more aggressive tumor behavior.

The univariate survival analyses (Table 3) revealed highly significant prognostic values for pT status, Fuhrman grade, nodal status, metastasis, and performance status (each P < 0.001). Age, gender, and histologic subtype were no prognosticators (each P > 0.100). Positivity for claudin-1 was significantly (P = 0.008) associated with shortened patient survival. This relationship could also be shown for clinically nonmetastasized tumors (P = 0.028; Fig. 2A). For the subgroup of clear cell renal cell carcinoma (n = 278) this prognostic value could be confirmed (P = 0.009; Fig. 2B) and was also relevant in the subgroups of nonmetastasized and asymptomatic patients (performance status, 0) with clear cell renal cell carcinoma (P = 0.027 and P = 0.048).

For the supplementary papillary renal cell carcinoma cohort there was a trend for shortened disease-specific survival times for claudin-1–negative tumors (P = 0.064). It should be considered, however, that the number of claudin-1 negative cases (n = 6; 4 events) was quite small (Fig. 2C).

To test for independence of the prognostic value of the different parameters a multivariate Cox regression analysis as done including Fuhrman grade, pT-status, metastasis, performance status, and claudin-1 expression. The nodal status was excluded from this analysis because its inclusion would have relevantly reduced the number of cases in the analyses.

In this Cox analysis claudin-1 lost its significance with P values of 0.061 for all histologic subtypes and 0.095 if only clear cell renal cell carcinomas were considered (Table 4). However, for the subgroup of patients with a performance status of 0 (asymptomatic), claudin-1 remained a multivariate significant prognosticator of patient survival for all histologic subtypes (P = 0.045; relative risk, 1.849; n = 216) and for those with clear cell renal cell carcinoma (P = 0.033; relative risk, 2.074; n = 189). Additionally, the same could be shown for the clinically relevant subgroup of nonmetastasized patients (P = 0.028; relative risk, 1.662; and n = 292 for all histologic types, and P = 0.030; relative risk, 1.745; and n = 254 for clear cell renal cell carcinoma only).

Conventional prognosticators pT status, Fuhrman grade, existence/absence of metastasis, and performance status remained almost constantly significant markers for disease-specific patient survival in these analyses.

Claudins have been selected as candidate markers for detection, prognostic evaluation, and therapy of various human cancers (15). In this study, by carefully analyzing a large and thoroughly characterized renal cell carcinoma cohort, we showed that claudin-1 expression is absent in majority of clear cell renal carcinomas. In contrast to clear cell renal cell carcinoma, claudin-1 was almost diagnostic for the papillary subtype of renal cell carcinoma, which was also confirmed on the supplementary tissue microarray. Because only very few chromophobe renal cell carcinomas were included, further studies are needed to evaluate the characteristics of the claudin-1 expression for these tumors.

Interestingly, claudin-1 protein expression was significantly associated with shortened disease-specific patient survival times in our main study cohort. The clinicopathologic differences between the clear cell and the papillary subtypes of renal cell carcinoma in our main study cohort seem not to be sufficient to explain these findings and the associations of claudin-1 expression with parameters of unfavorable tumor behavior. Especially for the subgroup of clear cell renal cell carcinoma, these results could be confirmed. Additionally, in univariate and multivariate analyses the presence of claudin-1 in clear cell renal cell carcinoma in asymptomatic patients (Eastern Cooperative Oncology Group performance status of 0) or patients without distant metastasis (M₀) allows for a more precise discrimination of tumors with distinct clinical behavior.

On the other hand, loss of claudin-1 expression in papillary renal cell carcinoma was more often found in more aggressive tumors and in patients with shortened disease-specific survival times. These results should be interpreted with caution due to the small number of cases in this group. However, the overall notion that claudin-1 is differentially expressed in clear cell and papillary renal cell carcinomas with different implications of up- and down-regulation of this protein, seems reasonable.

Another surprising result was the association of claudin-1 expression with male gender in clear cell and papillary renal cell carcinomas, without gender being a prognostic marker in any of the groups. Renal cell carcinomas are more often found in men than in women, a fact also deducible from our study cohorts. Nonetheless, we could not find any explanation for this phenomenon in the published literature on claudins.

The localization of claudin-1 in the Bowman's capsule and collecting duct systems is in line with previous studies (8, 39). Interestingly, claudin-1 expression seems to be reduced in prostate, breast, and thyroid cancer (18, 20, 36). Even a prognostic value of such a loss has been reported. Likewise, loss of claudin-1 correlated with malignancy of hepatocellular carcinoma (16). Previous studies underline that the loss of cell junction proteins precede epithelial-mesenchymal transition, an early process during cancer progression (40, 41). The transcription factors Slug and Snail play a crucial role in epithelial-mesenchymal transition and repress claudin-1 expression (42).

For colorectal cancer the existent data on claudin-1 are inconclusive concerning up- or down-regulation (14, 35). A recent study from Dhawan et al. found claudin-1 expression associated with metastasis and thus a more serious prognosis in colorectal cancer (12), which would match our findings in renal cell cancer.

The absence of claudin-1 in the majority of clear cell carcinomas could be seen as supportive of the theory that clear cell renal cell carcinomas commonly express markers that are immunophenotypical of proximal tubule epithelia, which intrinsically lack claudin-1. In this context, the abnormal expression of claudin-1 in this renal cell carcinoma subtype might be interpreted as a sign of further tumor dedifferentiation, which could explain the worse prognosis of these tumors. This is an assumption that is supported by the observation that there was a significant association between higher Fuhrman grade and claudin-1 expression in clear cell renal cell carcinoma.

It is notable that most papillary renal cell carcinomas, which are also thought to differentiate along the profile of proximal tubules, were positive for claudin-1. This might question the validity of the classic histiogenesis hypothesis for renal cell carcinoma.

Whether the aberrantly expressed claudin-1 proteins still fulfill a function similar to the physiologic tight junction formation or whether the expression of claudin-1 implicates any direct tumorbiological benefits for clear cell renal cell carcinoma remains to be studied. In fact, down-regulation as well as up-regulation of claudins in cancer may be attributed to functions unrelated to tight junction (28). Whereas overexpression may affect cell signaling pathways, down-regulation could lead to defects in cell adhesion that in turn abets the progression into invasive or metastatic neoplasm with local invasion and metastatic dissemination.

The gene encoding for claudin-1 is located on the long arm of chromosome 3. Alterations of this chromosome, mainly deletions and translocations in the short arm, are frequently found in familial and sporadic clear cell renal cell carcinoma (43, 44). In contrast, in papillary renal cell carcinoma trisomies of the long arm of chromosome 3 seem to be typical (45).

As stated above, the loss of claudin-1 seemed to be of prognostic value in some carcinomas but also the opposite was described for other tumor entities. For example, up-regulation of claudin-1 contributed to increased cell motility and invasiveness in squamous cell carcinomas and melanomas (17, 19).

The observation of diffuse cytosolic immunostaining of claudin-1 may suggest additional unexpected roles of claudin at cystosolic sites. Studies reported for other claudins that distribution is not restricted to tight junctions only, although the function of subcellular localizations remains unclear (2, 39). Even nuclear localization was reported (12). In this study we observed an additional nuclear staining in only very few cases.

Structurally, claudin-7 is the closest relative to claudin-1 which has also been shown to be differentially expressed (up- as well as down-regulated) in gastric, prostatic, esophageal, and colorectal cancer, and the chromophobe subtype of renal cancer (25, 46). For claudin-7, however, only down-regulation was associated with adverse tumor behavior (20, 47).

Renal cell carcinomas are characterized by resistance to chemotherapy and radiation with high morbidity and mortality. For decades no improvement in survival was seen. The discovery of claudin alterations and their prognostic effect in renal carcinoma underscores the pathophysiologic link between tight junction proteins and renal cancer, offering potential for new therapeutic approaches that are warranted. Experiments showed noteworthy therapeutic effects through the application of blocking antibodies and small interfering RNA–mediated knockdown of claudin-1 in hepatitis where claudin-1 serves as host viral coreceptor (48). Analogous mechanisms could be helpful in future cancer chemotherapeutics. In addition, pharmacologic molecules that modulate claudin were proposed to improve drug delivery (49).

In conclusion this study showed the prognostic value of claudin-1 in renal cell carcinomas (especially clear cell renal cell carcinoma) for univariate and multivariate conditions. Especially for asymptomatic patients and those with nonmetastasized clear cell renal cell carcinoma, claudin-1 expression could delineate additional prognostic information. Up-regulation of claudin-1 was also associated with dedifferentiation in clear cell renal cell carcinoma. Additionally, claudin-1 expression seems to be a diagnostic marker for papillary renal cell carcinoma, in which rather the loss and not the up-regulation is a hint of tumor dedifferentiation and aggressiveness. These interesting findings could be useful in the diagnostic, prognostic, and, given the proposed potential of the claudin family as therapeutic targets, predictive assessment of renal cancers.

No potential conflicts of interest were disclosed.

We thank Britta Beyer and Silvia Behnke for excellent technical assistance and the Sonnenfeld Stiftung for granting the tissue microarrayer.