Ruediger E. Port 1,4, C. Andrew Boswell 1, Annie Ogasawara 1, Herman Gill 1, Jan Marik 1, Alice F. Tarantal 2,3, Eric Berg 2, Simon P. Williams 1, Gregory Z. Ferl 1
1 Genentech (South San Francisco, USA), 2 University of California Davis (Davis, USA), 3 California National Primate Research Center (Davis, USA), 4 Consultant (Saarbruecken, Germany)
Introduction: The success of checkpoint-inhibition immunotherapy in cancer patients hinges on the presence of CD8+ (“cytotoxic”) T cells in tumors, and a variety of radiolabeled anti-CD8 PET tracers are currently being developed for evaluating individual chances of success by non-invasively detecting binding to CD8
in tumors[1-5]. The radioactivity detected in tumor, however, represents bound and unbound tracer so that it might reflect binding to CD8 as well as distribution of unbound tracer in the absence of CD8. The distinction of specific binding and non-specific distribution of PET tracers in tissues has successfully been achieved in the development of CNS drugs, typically by analyzing tracer kinetics in a brain region containing the target receptors, and in a “reference” region devoid of the receptors with the same parameters of non-specific tracer distribution[6,7]. No such reference tissue exists for individual malignant tumors. Instead, we are developing a non-binding control tracer, [18F]-2C8v145 (“C”) along with a structurally almost identical anti-CD8 tracer [18F]-2C8v144 (“B”), in order to monitor both tracers sequentially in blood and tumor and analyze kinetics in tumor by assuming the same parameters of non-specific distribution for both tracers[8-10].
Objectives: 1) Estimate the non-specific systemic PK parameters of C and B, and the parameters of B binding to CD8 in blood and elsewhere in the body, as required for estimating individual plasma concentration-time courses of C and unbound B, i.e. the vascular input functions needed for analyzing kinetics in tumors.
2) Predict total body radioactivity exposure and receptor (CD8) occupancy for arbitrary doses of B.
Methods: Each of three rhesus monkeys underwent [18F]-PET two times to monitor first C and, three days later, B in arterial blood for one hour. Mass doses/kg of each tracer varied 5-fold.
C and B are hydrophilic VHH peptides with MW = 14 kDa. B binds monovalently to CD8-alpha without cellular internalization of receptor-ligand complex[8,9]. The in-vitro equilibrium dissociation constant (K_D) of B with CD8 is 0.13 nM, as determined by SPR[9]).
Plasma concentrations were calculated using individual hematocrit, assuming no non-specific cellular uptake within one hour. Kinetics of C were described with a linear three-compartment model. Binding of B to CD8 was described by adding two compartments containing the amounts of CD8 within and outside of blood (R_b = R_tb*fR_b and R_4 = R_tb*(1 – fRb) where R_tb is the total body amount of CD8 and fR_b is the fraction present in blood), and the amounts of bound B in plasma and outside of blood (A_p.b(t), A_4(t)) linked to the unbound plasma concentration by two pairs of k_on/k_off rate constants (k_on.p, k_off.p and k_on.4, k_off.4). k_on/k_off = K_D was assumed to be the same throughout the body. k_on.p had to be fixed as estimated by SPR (0.50/(nM*min)) for model identifiability. NONMEM, version 7.50[11] was used for model fitting. Graphical and statistical analyses were done in R[12].
Results: The “typical” mean residence time of C, estimated for body weight = 2 kg, was 26 min. The estimated mean parameters of B were: TV(R_tb) = 3.4 nmol/2 kg, TV(fR_b) = 0.027, TV(k_on.p) (fixed) = 0.50/(nM*min), TV(k_on.4) = 0.00975/(nM*min) (more than 50 times less than TV(k_on.p)!), K_D = 0.23 nM[10]. Predicted total body receptor occupancy (RO_tb(t)) by B upon a 20 second i.v. infusion of 16, 4, 1 nmol/2 kg was still higher than 90%, 70%, 20% at one hour after the beginning of the infusion, when unbound B in plasma had dropped from the maximum at the end of the infusion by more than one order of magnitude[10].
Conclusions:
1) Testing C and B sequentially in the same animals enabled estimates of the “typical” amounts of CD8 within and outside of circulating blood, and predictions of total body receptor (CD8) occupancy by B versus time for arbitrary mass doses.
2) The similarity of the in-vivo and in-vitro K_D estimates for B suggests that the rate-limiting step behind the apparent slow down of association in vivo outside the circulation, compared to blood, is the same in the forward and backward directions.
3) Mixed-effects modeling of the plasma kinetics of C and B provided the population parameter estimates required for estimating individual plasma concentration-time courses of C and B, i.e. the vascular input functions needed for analyzing kinetics in tumor.
References:
[1] Farwell MD et al. J Nucl Med (2022) 63, 720-726.
[2] de Ruijter KL et al. Nat Med (2022), 28, 2601-2610.
[3] Tavaré R et al. Cancer Immunol Res (2022) 10, 1190-1209.
[4] Wang Y et al. Eur J Nucl Med Mol Imaging (2022) 49, 4394-4405.
[5] De Groof TWM et al. Eur J Nucl Med Mol Imaging (2024) 52, 193-207.
[6] Innis R et al. J Cereb Blood Flow Metab (2007), 27, 1533-1539.
[7] Varnaes K et al. J Pharmacokin Pharmacodyn (2013) 40, 267-279.
[8] Sriraman S et al. Eur J Nucl Med Mol Imaging (2023), 50, 679-691.
[9] Sriraman S et al. manuscript in preparation.
[10] Port et al. J Pharmacokin Pharmacodyn (2026), in review.
[11] [https://nonmem.iconplc.com/nonmem750] .
[12] https://www.R-project.org.
Reference: PAGE 34 (2026) Abstr 12204 [www.page-meeting.org/?abstract=12204]
Poster: Drug/Disease Modelling - Oncology