Re-Evaluating E-Cadherin and β-Catenin: A Pan-Cancer Proteomic Approach with an Emphasis on Breast Cancer

Abstract

E-cadherin is conventionally considered to be a good prognostic marker in cancer. The loss of E-cadherin is one of the key hallmarks of epithelial-to-mesenchymal transition, a biological process that promotes cancer cell invasiveness and metastasis. Recent evidence has cast doubt on the importance of epithelial-to-mesenchymal transition in metastasis. The availability of protein-level data in the Cancer Genome Atlas allows for the quantitative analysis of protein and prognosis. The prognostic values of E-cadherin and β-catenin were revisited across 19 cancer types, and high E-cadherin was found to correlate with good prognosis in most cancers. Conversely, higher E-cadherin and β-catenin correlated with shorter survival in invasive breast carcinoma. Stratifying breast cancers by histologic subtype revealed that the poor prognosis of E-cadherin and β-catenin proteins was characteristic of infiltrating ductal, but not lobular, carcinomas. To further corroborate the protein findings and examine cellular localization, immunohistochemistry was used for E-cadherin and β-catenin in 163 breast patient samples from the Iowa cohort. Most previous studies showing that reduced or absent E-cadherin and β-catenin was inversely associated with tumor stages in ductal carcinomas were confirmed. Taken together, these results lead us to question the prognostic values of E-cadherin and β-catenin in ductal carcinomas and indicate a complicated role of E-cadherin and β-catenin in breast cancer progression.

Figure 1

Protein level and prognostic value of β-catenin and E-cadherin across 19 cancer types. A: Z-score protein level of β-catenin and E-cadherin in The Cancer Genome Atlas Pan-Cancer cohort (P < 0.001 for β-catenin and E-cadherin). B: Summary of Cox proportional hazard regression comparing the upper quartile with the lowest quartile of β-catenin protein level. Dotted line indicates P = 0.05. C: Summary of Cox proportional hazard regression comparing the upper quartile with the lowest quartile of E-cadherin protein level. Dotted line indicates P = 0.05. Size of points is on a relative scale on the basis of –log10(P value). n = 46 [A, adrenocortical carcinoma (ACC)]; n = 127 [A, bladder urothelial carcinoma (BLCA)]; n = 820 [A–C, breast invasive carcinoma (BRCA)]; n = 326 [A, colon adenocarcinoma (COAD)]; n = 201 [A and B, glioblastoma multiforme (GBM)]; n = 203 (A and B, head and neck squamous cell carcinoma (HNSC)]; n = 444 [A–C, kidney clear cell carcinoma (KIRC)]; n = 207 [A, kidney papillary cell carcinoma (KIRP) and skin cutaneous melanoma (SKCM), and C, KIRP]; n = 257 [A, lower-grade glioma (LGG)]; n = 233 [A, lung adenocarcinoma (LUAD)]; n = 192 [A, lung squamous cell carcinoma (LUSC)]; n = 408 [A, ovarian serous cystadenocarcinoma (OV)]; n = 105 [A, pancreatic adenocarcinoma (PAAD)]; n = 164 [A, prostate adenocarcinoma (PRAD)]; n = 127 [A, rectum adenocarcinoma (READ)]; n = 299 [A, stomach adenocarcinoma (STAD)]; n = 374 [A, thyroid carcinoma (THCA)]; n = 404 [A and C, uterine corpus endometrial carcinoma (UCEC)].

Figure 2

The dynamics of β-catenin and E-cadherin in breast invasive carcinoma. A: Heat map of the clinical and pathologic features of The Cancer Genome Atlas invasive breast carcinoma (BRCA) cohort. Samples across the BRCA cohort underwent unsupervised clustering on the basis of the indicated protein levels. B: Density-based histogram and corresponding boxplot of β-catenin protein level by histologic diagnosis. Adjusted P < 0.001. C: Density-based histogram and corresponding boxplot of E-cadherin protein level by histologic diagnosis. Adjusted P < 0.001. D: Kaplan-Meier curve using β-catenin protein level to split infiltrating ductal carcinoma (IDC) samples into quartiles, using Cox proportional hazard regression comparing the upper quartile with the lowest quartile. E: Kaplan-Meier curve using E-cadherin protein level to split IDC samples into quartiles, using Cox proportional hazard regression comparing the upper quartile with the lowest quartile. F: Kaplan-Meier curve using β-catenin protein level to split infiltrating lobular carcinoma (ILC) samples into quartiles, using Cox proportional hazard regression comparing the upper quartile with the lowest quartile. G: Kaplan-Meier curve using E-cadherin protein level to split ILC samples into quartiles, using Cox proportional hazard regression comparing the upper quartile with the lowest quartile. n = 873 (A); n = 631 (B, IDC); n = 152 (B, ILC); n = 6 (B, medullary carcinoma); n = 23 (B, mixed histology); n = 60 (B, other). ^∗P < 0.05, ^∗∗∗∗P < 0.0001. DVL, segment polarity protein dishevelled homolog 3; ER, estrogen receptor; GSK, glycogen synthase kinase; HER2, human epidermal growth factor receptor 2; HR, hazard ratio; NA, not applicable; PR, progesterone receptor.

Figure 3

Immunohistochemical (IHC) staining results for β-catenin and E-cadherin in breast cancer samples. Representative IHC staining patterns that are observed: membranous staining for both β-catenin (A) and E-cadherin (B); reduced membranous expression for both β-catenin (C) and E-cadherin (D); absence of staining for both β-catenin (E) and E-cadherin (F); and membranous and nuclear staining of β-catenin (G) with strong E-cadherin membranous staining (H). Arrows indicate cells with nuclear staining of β-catenin. Original magnification, ×400 (A–H).

Figure 4

β-Catenin and E-cadherin protein in infiltrating ductal carcinoma (IDC) samples on the basis of immunohistochemical designation. Outliers, defined as E-cadherin <−3 and β-catenin >−3, circled withdotted line. A: Scatterplot of β-catenin and E-cadherin protein across IDC samples. B: Density histogram of mean-centralized quotient of β-catenin and E-cadherin. Cutoff values of −0.65 (low, left-bound dotted line) and 0.8 (high, right-bound dotted line) were used on the basis of distinct subpopulations. χ² Tests were performed comparing low, high, and unaffected population distributions. C: Kaplan-Meier curve from the estrogen receptor (ER)⁺ (blue), human epidermal growth factor receptor 2 (HER2)⁺ (green), and triple-negative breast cancer (TNBC; red) samples in the METABRIC cohort using CDH1 levels split by auto-cutoff for high versus low samples, using Cox proportional hazard regression. NA, not applicable.