Helical CT data were acquired for anatomical reference and for attenuation correction using Nucline (V2.01, Mediso). Images were acquired with a semicircular scan, with 180 projections. The X-ray energy was set to 55 kVp with 600 milliseconds exposure and 1:4 image binning. Images were reconstructed using Butterworth filtering with an isotropic voxel size of 213 μm. PET images, with a nominal isotropic resolution of 0.6 mm, were reconstructed using a dynamic protocol (Tera-Tomo 3D (Mediso) algorithm), energy window 400–600 keV, coincidence modes 1–5, full detector model, normal regularization, 2 iterations, 6 subsets, with attenuation and random scatter correction. Images were an-alyzed in VivoQuant® software (vs. 4.0 patch 1, InviCRO, Needham, Massachusetts, USA). A dynamic PET acquisition lasting 120 min was initiated 30 s prior to intravenous injection of 3.7 ± 1.2 MBq 68Ga-C2Am (1075 ± 382 μg protein/kg body weight; 10 mL/kg; SA = 0.15 ± 0.05 MBq/μg protein). Scans were reconstructed into 23 time bins (4 between 0–1 min; 4 between 1–5 min; 11 between 5–60 min and 4 between 60–120 min).

Quality control consisted of ultra-performance liquid chromatography using a Superdex 75 Increase 5/150 gel filtration column (Cytiva, Emeryville, California, USA), which is designed for separating low molecular weight proteins (3 kDa to 70 kDa). Small molecules, including [68Ga]GaCl3, have long retention times (>10 min).

Correlation analysis of tumor PET signal with the histological marker of apoptosis, CC3 staining (%) was also performed and raw staining images are attached here.