Tumor samples
Neuroblastic formalin-fixed paraffin-embedded (FFPE) blocks from 2004–2014 were obtained from the Pequeno Príncipe Hospital (PPH) Biobank, and the clinical data were retrospectively collected from the PPH Service of Medical Archives and Statistics (SAME). This study was approved by the PPH Ethics Committee (CEP0518-07/2007; amendment 2014), and all patients were kept anonymous. Two subgroups of NB specimens were obtained, 82 from patients that were not submitted to any treatment (pre-CT), and 20 were obtained after chemotherapy (post-CT). In eleven patients, it was possible to obtain tumor specimens pre- and post-CT. Each sample was classified according to the following criteria: (i) age at diagnosis (< 18 or ≥ 18 months); (ii) sex; (iii) Shimada classification 20 (favorable or unfavorable histological features); (iv) INSS-based staging (1–4 or 4S); and (v) clinical course of the disease (alive without disease, relapsed, deceased).

Tissue microarray (TMA) construction and immunohistochemical analysis

Ten TMAs were sampled, each of them with an average of two cores (4 µm thick; 3 mm diameter) from 12 cases 21. The TMAs were incubated with a primary rabbit polyclonal anti-COX-2 (1:200; Spring Bioscience, USA) and secondary–horseradish peroxidase (HRP)–conjugated antibody (Spring Bioscience USA). Positive (colon cancer specimens) and negative (omitting primary antibodies) controls were run in parallel in each of the reactions. The images were obtained using a Zeiss Axioscan Slide Scanner (Jena, Germany) in high power fields (HPFs; 20x magnification), with a total area of 90,472.78 µm2. Immunohistochemical expression was evaluated through quantitative analyses of cytoplasmic staining images using the Image-Pro Plus® software (Rockville, MD, USA) and calculated as a percentage of the ratio of positive staining area per the total area22.