Page 29 - 89Zr-Immuno-PET:Towards a Clinical Tool to Guide Antibody-based Therapy in Cancer

Dosimetry of immuno-PET with 89Zr-cmAb U36
Validation of quantitative 89Zr-immuno-PET imaging
OSEM reconstructions with attenuation correction of the immuno-PET scans
were used to investigate the accuracy of quantification 89Zr-cmAb U36 PET in
vivo. To this end, three 1.5 cm diameter ‘regions of interest’ (ROIs) were defined 2 within the left ventricle of the heart in three subsequent image planes using an
axial view of the PET images. The average activity concentration within these
ROIs, representing PET assessed bloodpool activity concentration, were directly
compared with the activity concentration measured in manual blood samples
(“sampled blood activity”) using a calibrated well counter in order to validate the quantitative accuracy of 89Zr-PET in vivo.
In addition, the uptake of the radioimmunoconjugate in tumors (%ID/ cm3) as assessed from scans acquired at 144 h p.i., was compared with the uptake (%ID/g) derived from tumor biopsies collected at 168 h p.i. %ID/cm3 as derived from the PET scans was converted to %ID/g using a soft tissue density factor of 1.04 g/cm3.
VOI definition
The uptake in different organs was determined by manually defined ROIs using the Clinical Application Programming Package (CAPP, provided with the ECAT software). 89Zr uptake was determined in the following organs: lung, liver, spleen, kidney, heart, and if visible, thyroid. In addition, a semi-automatic ‘volume of interest’ (VOI) was defined over the tumor using in-home developed software tools (22).
Details of ROI/VOI definition: As most organs could be visually best identified on the 1 h p.i. emission scans, most organ 3D VOIs were defined manually using this early uptake scan. However, VOIs for the thyroid and tumor were defined on the last scan, because their delineation was optimal at this last time point. Finally, lung regions were defined on reconstructed transmission scans, using the semi-automatic isocontour tool within the CAPP software. All 3D VOIs were defined using a coronal view of the PET images.
A 3D VOI was generated from multiple 2D ROIs by grouping these ROIs into the VOI. The ROIs on the first and last slices were not used for the VOIs, since these were prone to partial volume effects (spill-in and spill-out) and/or sampling errors. The VOIs defined on the 1 h p.i. scan were saved and imported to the other time frames, where the organs were not as clearly visible. To allow for projection of VOIs onto scans obtained at other time intervals, patients were scanned in the
27