*2.2. IOUS and Intraoperative Navigation*

At the point of laparotomy, intraoperative ultrasonography (IOUS) was performed using an Esaote Twice ultrasound system (Esaote, Genoa, Italy), equipped with an intraoperative T-shaped probe (IOT332 probe, Esaote, Genoa, Italy), working at 3–11 MHz frequency.

The two imaging modalities (PET-CT and IOUS) were synchronised by a semi-automatic system in the following steps. The images of PET-CT were uploaded to the ultrasound machine and then projected onto the screen beside the standard IOUS images. Some intrahepatic anatomic landmarks identified on the PET-CT (e.g., umbilical portion, firstorder bifurcation of the right portal branch, hepato-caval confluence) were manually identified at IOUS. Once a landmark was visualised at IOUS, the axial image of PET-CT with the same landmark was identified and selected, and a mark was placed on the anatomical structure in the two imaging modalities. After the identification and selection of two landmarks, the machine provided an automatic synchronisation of the two imaging modalities. The correct synchronisation between the two imaging modalities was then verified by scanning all the liver. If any discordance was observed, some additional

anatomic landmarks were selected to refine the process until a perfect overlapping of all intrahepatic anatomical structures was achieved. Once the process was completed, the overlapping of the full liver was obtained, and the PET-CT navigation was possible (Figure 1). The SUVmax and the SUVmin tumour areas selected at PET-CT were identified and sampled using a 16-gauge Trucut needle (Figure 2); the needle trajectory was selected to be fully included within the resected portion of the liver to avoid any risk of tumourseeding in the future liver remnant. At least two biopsies were taken from each area. At the end of the resection, the same targeted areas were again identified, and a macrobiopsy was performed. Only samples with an adequate cellular composition were retained for the analyses (through a quick histologic check after sampling).

**Figure 1.** Navigation technology with the intraoperative fusion of preoperative PET-CT and IOUS. The tumour areas having different uptake at PET-CT are identified in vivo during surgery. (**a**) PET axial view of ICC; (**b**) Intraoperative fusion of PET-CT and IOUS images.

**Figure 2.** IOUS-guided biopsy of the tumour areas having different uptake at PET-CT.

#### *2.3. Pathology Analyses*

The specimens were fixed in formalin, paraffin-embedded, and stained with haematoxylineosin. Each sample had a standard morphological evaluation. Immunohistochemistry (IHC) was used to analyse the following parameters: the expression of CK7 and CK19; immune infiltrate (CD3+, T-lymphocyte marker; CD4+, helper/inducer T-lymphocyte marker; CD8+, suppressor/cytotoxic T-lymphocyte marker; CD68+, macrophage marker; and CD163+, M2 macrophage marker); the expression of programmed cell death protein 1 (PD1), its ligand (PD-L1), and tumour protein p53; proliferation index (Ki67); and metabolic enzymes glucose-6-phosphate dehydrogenase (G6PD) and citrate synthase (CS). We analysed the presence of FGFR2 translocations and the presence of microsatellite instability, and the loss of heterozygosity (1p36) using fluorescence in situ hybridisation. For PD1, PD-L1, p53, Ki67, G6PD, and CS, data were expressed as the percentage of immunoreactive cells compared to the total number of neoplastic cells. For the immune infiltrate (CD3, CD4, CD8, CD68, CD163), data were expressed as the percentage of immunoreactive cells compared to the total number of immune cells.

The specimens from SUVmax and SUVmin areas were separately analysed, and their data were compared. Both IOUS-guided tumour biopsies before resection and macroscopical biopsies at the end of resection were analysed. The concordance between samples from the same area was assessed.
