Autostainer

The anti-proliferative activity of test formulations was determined by tumor immunohistochemical staining of the Ki-67 proliferating protein in formalin-fixed, paraffin-embedded specimens (Yi et al., 2018 (link)). Briefly, 4 μm-thick tumor tissue sections were dried, deparaffinized, and rehydrated following standard procedures. Sections were subjected to heat-induced antigen retrieval. Immunohistochemical staining was performed using Ki-67 antibody (Monoclonal Mouse Anti-Human, Dako Agilent autostainer (Twinsburg, OH)). Optical densities of the Ki 67-positive areas in the tumor sections were measured and the % Ki-67 proliferative protein expression calculated. This was achieved by digital image analysis using Image J software (version 1.45s, Bethesda, MD) together with computer-assisted microscopy (El Sayed et al., 2018 (link)).

Multi-tissue arrays (TMA) containing two 1.5-mm representative punches of 62 TH and TC patient samples were used for immunohistochemical (IHC) stainings (Table 1). All antibodies were established on positive control tissues chosen from the Human Protein Atlas (http://www.proteinatlas.org). Stainings were performed on 2-μm sections according to a standard protocol on an Autostainer (Agilent, USA). In brief, antigen retrieval was performed at 95°C in pH 6 or pH 9 Envision FLEX target retrieval solution in a PT Link Module (Agilent, USA) followed by 1-h incubation with primary antibodies (Additional file: Table S2). Samples were washed with PBS and incubated with an appropriate secondary antibody (EnVision Flex+, Dako) for 30min. Two individual observers (DM and PS) evaluated stainings for both cores of a respective case and graded as positive when >50% of the tumor cells were positive. Staining intensity was scored 0 to 2 (0, negative staining; 1, weak staining; 2, strong staining). The average staining intensity of two cores was taken for further analysis. To evaluate the best clinical separation and to define the optimal threshold for dividing IHC low and high staining intensities, the cutoff finder was used as described by Budczies et al. [33 (link)]. These resulting cutoff values are given in the corresponding figure legends.