Improved mehtods and reagents for pretargeted radioimmunotherapy of cancer

Abstract

Pretargeted radioimmunotherapy (PRIT) of cancer improves upon conventional radioimmunotherapy (RIT) by decoupling the pharmacokinetics of the targeting agent and the radioisotope. In order to improve upon PRIT, we have considered variables such as treatment setting and methodology, the transport and clearance characteristics of targeting agents, and the radionuclides used for therapy. PRIT has been modeled with the aim of examining the theoretical potential of PRIT under optimal conditions to kill every cell in malignant, avascular micrometastases. A mathematical model of PRIT was developed that combined a two-compartment pharmacokinetic model, antibody binding kinetics, diffusion and catabolism in tumor spheroids, and radiation dosimetry models for alpha- and beta-emitting radionuclides. This model demonstrated that it is theoretically possible to kill every cell in 100 tm radius micrometastases using 9Y- or 213Bi-based PRIT with acceptable toxicity as described. The therapeutic window for dosing radionuclide-carrying hapten was found to be strongly dependent on cell-specific parameters such as antigen concentration, void fraction, and the radiosensitivity parameter a, as well as on targeting agent molecular parameters such as the diffusivity and antigen-binding association rate.

(cont.) Surprisingly, the therapeutic window was insensitive to the radiosensitivity metric a/I, the targeting agent antigen-binding dissociation rate, and all pharmacokinetic parameters. Overall, 213Bi-based PRIT significantly outperformed 9Y-based PRIT in terms of the safe therapeutic time window for radiometal dosing and the degree of cell overkill that could be achieved. An attempt was made to isolate high-affinity scFv or linear peptide binders against the loaded metal chelate Ga-DOTA-biotin. Unfortunately, several different approaches led only to scFvs and linear peptides with at best micromolar affinity for Ga-DOTA-biotin. It is possible that Ga-DOTA-biotin is a difficult target against which to engineer high affinity binders due to the chelate's six-coordinate binding of the gallium ion, which may result in rapid exchange of the carboxyl arms of the chelate in solution. As an alternative approach to targeting agent design, an anti-CEA, anti-fluorescein single-chain bispecific diabody was designed, produced in S. cerevisiae and characterized. The full-length diabody (55 kDa) binds CEA expressed on the surface of colorectal cancer-derived SW1222 cells with a KI of 4.3 ± 2.5 nM, and also binds fluorescein while bound to CEA on the cell surface.

(cont.) Lastly, in order to assist in protein engineering via directed evolution, asymptotically optimal probability estimation was combined with numerical bootstrapping and non-linear curve fitting to make accurate predictions of the actual underlying diversities of populations based on small samples of data.