Quantification of ⁸⁹Zr-immunoPET: Methodological considerations.

PURPOSE: Clinical 89Zr-immunoPET relies on semi-quantitative metrics, such as standardised uptake values (SUV) from static imaging at delayed time points. However, SUV can misinterpret target engagement as it ignores plasma and tissue kinetics. Kinetic modelling, common for shorter-lived tracers, is rarely applied to 89Zr-labelled monoclonal antibodies (mAbs) due to slow kinetics. This study characterises 89Zr-immunoPET kinetics using a two-tissue compartment model, identifies factors affecting tracer uptake, and assesses the validity of semi-quantitative metrics.

PROCEDURES: Realistic plasma input functions with different clearance rates, representing variable mAb kinetics and cold mAb doses, were derived from previous studies. Realistic ranges for kinetic rate constants and blood volume fractions were estimated based on literature. Each parameter was varied to assess its effect on 89Zr-immunoPET time-activity curves and Patlak analysis. SUV and target-to-plasma ratios (TPR) at delayed time points were compared with Patlak net irreversible rate of uptake (K i).

RESULTS: Fast plasma kinetics resulted in small but specific uptake, while slow kinetics produced higher signals with larger nonspecific contributions. SUV and K i were driven by tracer delivery (K 1) and residualisation upon binding (k 3), depending on the scenario. Within groups with limited variations in plasma clearance, semi-quantitative metrics correlated well with K i. For slow kinetics, where nonspecific uptake contributes more prominently, TPR performed better than SUV. However, despite good agreement in general, linear fits between semi-quantitative metrics and K i showed notable variability, indicating their limited reliability as surrogates for K i. Between groups with large differences in plasma clearance, only normalisation by the integral of the plasma input function, as in Patlak analysis, enabled accurate interpretation.

CONCLUSIONS: Patlak analysis accounts for plasma kinetics and reversible signals, enabling more accurate interpretation of 89Zr-immunoPET data than SUV. Wider use of kinetic modelling could enhance quantitative accuracy, support reinterpretation of past studies, and guide future study design.