Materials and Methods
Patients and Samples
Samples from 110 patients with GEP NETs were collected at the Hospital Universitario Central of Asturias and Hospital Virgen del Rocío of Seville, Spain, between February 2000 and October 2009, with institutional review board approval for guidelines on ethical procedures. The diagnosis of NETs was based on morphologic criteria according to the updated WHO grading classification in 2010. The characteristics of the studied patients (age, sex, and risk behaviors) and the clinicopathologic features of their tumors (WHO classification, tumor site and size, differentiation, stage, mitotic index, proliferation index, vascular invasion, lymph node status, necrosis, and the presence of metastasis) are shown in Table 1. The anatomic locations of the 110 tumors were the stomach (18), small intestine (28), pancreas (22), colon/rectum (23), and appendix (19).
Tissues obtained from biopsy specimens were fixed in 10% formaldehyde and paraffin embedded, cut 4 μm thick, mounted on treated slides, and stained with H&E. The neuroendocrine phenotype was confirmed by chromogranin A and synaptophysin immunohistochemical staining. Normal tissue present in each tissue section was considered a reference.
Tissue Microarray Construction
Representative tumor regions were identified on H&E-stained paraffin sections in each sample and selected to make 4 tissue microarrays containing 3 tissue cores from each of the 110 GEP NETs. Core cylinders 1 mm in diameter were processed using Microarrayer (Beecher Instruments, Sun Prairie, WI). Three core cylinders, punched from the donor block, were then deposited in the recipient paraffin block. After 5 minutes at 60°C, the tissue microarray blocks were cut in 4-μm-thick sections, ready for immunohistochemical techniques.
Immunohistochemistry
The automated system DISCOVERY (Ventana Medical Systems, Tucson, AZ) was used to carry out the immunohistochemical protein detection of interest. Sections were deparaffinized and rehydrated in EZ Prep (Ventana Medical Systems) for 20 minutes. Antigen retrieval was done by heating citrate buffer solution (pH 6.5) and HCl-Tris buffer solution (pH 9.0). Nonspecific antibody binding was blocked using casein (Antibody Block, Ventana Medical Systems) for 20 minutes. Endogenous peroxidase activity was blocked with H2O2 solution (Inhibitor, Ventana Medical Systems) for 4 minutes. Samples were incubated at 37°C with primary antibodies described in Table 2. The slides were incubated with the secondary antibody (OmniMap, Ventana Medical Systems) for 30 minutes at room temperature. Then the samples were visualized with 3,3'-diaminobenzidine (DAB, Ventana Medical Systems). Finally, samples were counterstained with hematoxylin (Ventana Medical Systems) and dehydrated and mounted in Entellan (Merck, Darmstadt, Germany). The sections were studied and photographed (20× objective) under a light microscope (Eclipse 80i, Nikon, Tokyo, Japan).
Double Immunohistochemistry
To assess the expression of EMT markers in healthy neuroendocrine cells, we performed double immunohistochemistry in normal intestinal mucosa, using synaptophysin antibody (MRQ-40, Ventana Medical Systems) with an ultra-View Universal Alkaline Phosphatase Red Detection Kit as chromogen (Ventana Medical Systems). The EMT markers were visualized with DAB as chromogen.
Immunohistochemistry Assessment
The protein expression levels were evaluated by a semi-quantitative method by 2 independent observers (A.A. and L.G.I., with a third observer [M.V.G.] in case of disagreement), taking into account 2 parameters: immunohistochemical signal intensity (graded on a 0- to 3-point scale) and the percentage of positive cells (0–100). All tumor cells in a core were considered to assess the percentage of positive tumor cells in a 20× field (1.2 mm). The product of both parameters rendered a score for each specimen. For statistical purposes, the tumors were divided into 2 groups, taking the median score value for each marker as a cutoff point.
For cadherins and β-catenin, their presence or absence in the cell membrane was recorded. A variable reflecting the integrity of the E-cadherin/β-catenin complex in the cellular membrane was created, including 2 categories: integrity retained (both molecules showing a membranous pattern) and integrity lost (expression of at least 1 that was not observed in the membrane). For EMT markers (Snail1, Snail2, Twist, and Foxc2), only the nuclear immunostaining signal was considered.
Statistical Analysis
The experimental results distributed among the different clinical groups of tumors were tested for significance employing the χ test (with Yates correction, when appropriate). Survival curves were calculated using the Kaplan-Meier product limit estimate. Deaths from causes other than the index tumor or its metastases were not considered treatment failures, and these patients were censored in all analyses involving length of survival. Differences between survival times were analyzed by the log-rank method. Multivariate Cox proportional hazards models (forward Wald method) were used to examine the relative impact of statistically significant variables in a univariate analysis. All statistical analysis was performed with SPSS 19.0 (SPSS, Chicago, IL). All tests were 2-sided. P values less than .05 were considered statistically significant.