Introduction
Upper tract urothelial carcinoma (UTUC) is a rare disease, accounting for 5–10 % of all urothelial carcinomas [
1]. In the last two decades, management of UTUC has improved but still remains challenging. Radical nephroureterectomy (RNU) remains the standard treatment for non-metastatic disease [
1,
2]. However, kidney-sparing approaches are now considered for low-risk UTUC, and conversely, regional lymphadenectomy and perioperative chemotherapy are discussed in addition to RNU for high-risk UTUC [
1,
2]. Current concerns lay in the identification of the patients who may benefit from these treatments. Recent evidence suggests that carcinogenetic mechanisms in UTUC are different from urothelial carcinoma of the bladder (UCB) [
2‐
4]. Therefore, molecular alterations from one setting may not be extrapolated in the other. In this regard, specific validation of biomarkers in UTUC is mandatory to develop predictive tools that could allow accurate clinical decision-making in the management of UTUC patients.
Decreased expression of the membrane-associated glycoprotein E-cadherin has been established as a feature of epithelial–mesenchymal transition (EMT) in epithelial malignancies [
5‐
7]. In normal cells, E-cadherins’ cytoplasmic domain binds with subtypes β or γ of the catenin proteins, which in turn secure attachment to the actin microfilament, thus ensuring cytoskeleton integrity and stable cellular adhesion [
8]. Loss of cellular adhesion is a tipping point in tumor progression resulting in poorly differentiated and invasive tumors [
7,
8]. E-cadherin has been shown to be an independent prognostic factor in UCB [
3,
9]. In UTUC, single-center studies with small cohorts have investigated the role of E-cadherin expression with conflicting results [
10‐
15]. We hypothesized that E-cadherin expression in RNU specimens was associated with features of biologically and clinically aggressive UTUC, thereby potentially helping in the clinical decision-making of UTUC patients. To assess this hypothesis, we tested the association of E-cadherin with pathologic characteristics and prognosis in a large multi-institutional cohort of patients treated by RNU for UTUC.
Discussion
In this study, we assessed the clinical significance of a decreased E-cadherin expression in an international cohort of 678 UTUC patients treated with RNU. We found that decreased E-cadherin expression in tumor cells is associated with adverse clinicopathological features and worse outcomes.
Half of the patients in this cohort presented with decreased expression of E-cadherin in the tumor. This was within the range previously reported in UCB patients (31–77 %) [
8,
9] but lower than that reported in UTUC patients (68–71 %) [
10,
11]. This could be due to our lower proportion of high stage tumors compared to the two other studies (50 vs. 54 and 61 %), as well as methodological differences in scoring, staining protocols, choice of antibody and/or antigen retrieval.
Patients with decreased E-cadherin were most likely to harbor tumors with features of biologically aggressive disease. This association is in line with the biological role of E-cadherin, as a calcium-dependent glycoprotein essential to epithelial tissue integrity. Loss of cellular adhesion results in the detachment of cancerous cells from the primary lesion, promoting invasiveness [
23]. In carcinoma in situ of the bladder, for example, loss of E-cadherin expression predicts RFS, disease progression and CSS [
21]. Similar results were reported in various UCB studies [
8,
9,
24‐
27] and one UTUC study [
10]: loss of E-cadherin immunoreactivity strongly correlated with advanced stage and high-grade tumors.
We further evaluated the relevance of E-cadherin as a biomarker to predict outcomes after RNU. The role of E-cadherin expression as a prognostic factor in urothelial carcinoma was supported by previous studies mainly focusing on UCB [
8,
9,
24‐
27]. Our results confirm that decreased E-cadherin expression is indeed associated with a higher probability of disease recurrence and cancer-specific mortality in UTUC. However, when adjusted for the effects of established prognostic factors in multivariable analyses, E-cadherin expression lost its independent prognostic value and, therefore, may have only limited value in clinical practice. Previous studies that addressed the relationship between E-cadherin and outcomes in UTUC led to conflicting results. Fromont et al. [
14] showed, in a cohort of 62 UTUC patients, that decreased E-cadherin expression was an independent prognostic factor for disease-free and overall survival. Conversely, most of the studies published thereafter with larger cohorts failed to demonstrate independent association between E-cadherin expression and disease recurrence after RNU [
10‐
13].
Consistent with the literature [
10‐
13], we found, in our subgroup analysis that, E-cadherin failed to demonstrate any independent prognostic value, outlining its strong association with other established pathological prognostic factors. We found a significant association between E-cadherin expression and adverse clinicopathological features such as advanced pathological tumor stage, high pathological tumor grade, lymph node metastases, LVI, concomitant carcinoma in situ, multifocality, tumor necrosis and sessile architecture. All these factors have been independently associated with worse outcome in UTUC [
1,
2,
19,
20,
28‐
30].
The biological and clinical roles of the E-cadherin-related pathways in urothelial carcinomas are yet to be understood. Indeed, the regulation of E-cadherin is linked to many different biomarkers [
2,
3]. Some of them have been already assessed in UTUC. Among them, Snail, a transcription factor is thought to repress the transcription of E-cadherin by binding to elements found in the E-cadherin promoter [
12]. An increased Snail expression has been reported as an independent prognostic predictor of recurrence-free and CSS [
12]. During EMT, it is thought that transcriptional regulation results in suppression of epithelial markers and gain of mesenchymal markers [
7,
12]. This process recognized in different types of cancer including bladder cancer [
31,
32] and UTUC [
11] has been observed between epithelial E-cadherin and mesenchymal N-cadherin and has been termed the “cadherin switch.” The novel genotype results in an alteration of normal tissue architecture and high-grade, invasive tumors. To our knowledge, only one study by Muramaki et al. [
11] addressed the role of N-cadherin expression in UTUC: In this study including 59 patients, N-cadherin expression was an independent prognostic factor of intra- and extra-vesical recurrence after RNU. Combining several biomarkers may help characterize the different pathways involved in tumor aggressiveness and create a prediction algorithm that would improve prognostication, clinical outcome and thus patient survival [
33]. At this time the ideal combination of biomarkers remains unfortunately elusive. From the bladder cancer literature and some preliminary upper urinary tract literature, cell cycle markers (p53, pRB, p21, p27 and cyclins), apoptosis markers (Fas, caspase-3, Bcl-2 and survivin) and proliferation markers (Ki67) may be used for a combined approach [
34]. Snail expression, N-cadherin expression, AKT pathway, β- or γ-catenins and matrix metalloproteinases are molecular markers associated with EMT that could be analyzed in a combined approach with E-cadherin expression.
We acknowledge that our study has some limitations. First and foremost are those related to its retrospective nature and the immunohistochemical technique. Indeed, this latter technique may be associated with a lack of reproducibility related to the choice of antibodies, the specimen handling procedures, technical demands and scoring protocols. However, tissue microarray with staining protocols and automated scoring systems based on bright field microscopy imaging coupled with advanced color detection software were used to overcome these common limitations. Finally, decreased expression of E-cadherin was defined according to a standard cutoff used in UCB and use of new thresholds may have led to different conclusions.
Acknowledgments
Open access funding provided by Medical University of Vienna.