*2.2. Immunohistochemical Analysis*

Histopathological analyses were performed on 3 μm sections of formalin-fixed paraffin-embedded sections of 27 tumors from 27 patients with newly diagnosed NF-PitNETs that were determined on the basis of the hormonal status in the peripheral blood. NF-PitNETs are usually soft and easy to remove via aspiration. A small amount of tissue was used for pathology assessment. In the present study, a large size of tissue was selected because the multiple, most vascularized regions (hot spots) should be screened for regionally averaged positive cell counts. Mitotic activity was assessed using hematoxylin and eosin (H&E) staining. Immunohistochemistry was performed according to standard procedures [20]. After tissue sections were deparaffinized and rehydrated, antigen retrieval was performed in citrate buffer (Ki-67, p53, VEGFR1, CD34, Foxp3, CD163, CD3, CD4, and PD-1), or in Tris buffer (pH 9 for VEGF-A, VEGFR2, CD8, HIF-1α, and PD-L1) using microwave irradiation or autoclave (HIF-1α and PD-L1). The sections were blocked for 60 min in 2.5% horse serum (ImmPRESSTM Detection Systems, Vectorlabs, CA, USA). The sections were incubated overnight at 4 °C with anti-Ki-67 antibody (1:200, M7249, DAKO), anti-p53 monoclonal antibody (1:100, DO-7, DAKO), anti-VEGF-A antibody (1:200, JH121, Merck Millipore), anti-VEGFR1 antibody (1:200, AF321, R&D SYSTEMS), anti-VEGFR2 antibody (1:600, 55B11, Cell Signaling Technology), anti-CD34 antibody (1:100,°C F1604, Nichirei Biosciences Inc.), anti-Foxp3 antibody (1:100, ab54501, Abcam), anti-CD163 antibody (1:100, ab87099, Abcam), anti-CD3 antibody (1:100, ab5690, Abcam), anti-CD4 antibody (1:200, 1F6, Nichirei Bioscience Inc.), anti-CD8 antibody (1:50, ab17147, Abcam), anti-hypoxia-inducible factor-1α (HIF-1α) antibody (1:100, H-206, Santa Cruz Biotechnology), anti-PD-1 antibody (1:50, NAT105, Abcam), and anti-PD-L1 antibody (1:500, 28-8, Abcam), then incubated with anti-mouse, anti-rabbit, or anti-goat

Ig secondary antibody (ImmPRESSTM Detection Systems, Vectorlabs) for 60 min at room temperature. The products were visualized with a peroxidase-diaminobenzidine reaction.

For the assessment of Ki-67 index, manual counting of 1000 tumor cells was routinely done at a high-power field (HPF: ×40) [21]. The positivity of VEGF-A staining in the tumor cytoplasm or stroma was assessed as the following: ++, diffuse intense staining; +, diffuse faint staining; −, negative staining. The staining positivity of VEGFR1 and VEGFR2 on endothelial cells was assessed as the following: +, staining in vascular endothelial cells; −, negative staining. For the assessment of microvessel density (MVD), the tissue sections were screened at low-power fields (×4), and the three most vascularized regions (hot spots) were selected for each region. The counting of microvessels was performed on these regions at HPFs (×20, 0.95 mm2). HIF-1<sup>α</sup> expression was assessed as the following: ++, expression in >10% of tumor cells; +, expression in ≤10% of tumor cells; −, negative staining [22]. For the assessment of density of Foxp3, CD163, CD4, and CD8 (+) cells, the tissue sections were screened using each immunohistochemistry at the low-power fields (×4), and three hot spots were selected. Counting of the positive cells was performed in these areas at the HPFs (×40, 0.47 mm2). PD-L1 expression was assessed as the following: 3+, expression in ≥50% of tumor cells; 2+, expression in ≥5% and <50% of tumor cells; 1+, expression in ≥1% and <5% of tumor cells; 0, expression in <1% of tumor cells [23]. Both histopathological reviewing and scoring were independently performed with blinded clinical information by three authors (MS, RT, and YM).

The specificity of immunohistochemistry was checked using negative and positive controls. For negative controls, paraffin sections were incubated with non-immune mouse, rabbit, and goat IgG at the same concentration used for each antibody. Sections from glioblastomas were used as the positive controls for each antibody (Figure S1).
