Drug-regulated CD33-targeted CAR T cells control AML using clinically optimized rapamycin dosing.

Chimeric antigen receptor (CAR) designs that incorporate pharmacologic control are desirable; however, designs suitable for clinical translation are needed. We designed a fully human, rapamycin-regulated drug product for targeting CD33+ tumors called dimerizaing agent-regulated immunoreceptor complex (DARIC33). T cell products demonstrated target-specific and rapamycin-dependent cytokine release, transcriptional responses, cytotoxicity, and in vivo antileukemic activity in the presence of as little as 1 nM rapamycin. Rapamycin withdrawal paused DARIC33-stimulated T cell effector functions, which were restored following reexposure to rapamycin, demonstrating reversible effector function control. While rapamycin-regulated DARIC33 T cells were highly sensitive to target antigen, CD34+ stem cell colony-forming capacity was not impacted. We benchmarked DARIC33 potency relative to CD19 CAR T cells to estimate a T cell dose for clinical testing. In addition, we integrated in vitro and preclinical in vivo drug concentration thresholds for off-on state transitions, as well as murine and human rapamycin pharmacokinetics, to estimate a clinically applicable rapamycin dosing schedule. A phase I DARIC33 trial has been initiated (PLAT-08, NCT05105152), with initial evidence of rapamycin-regulated T cell activation and antitumor impact. Our findings provide evidence that the DARIC platform exhibits sensitive regulation and potency needed for clinical application to other important immunotherapy targets.

GFAP and NfL increase during neurotoxicity from high baseline levels in pediatric CD19-CAR T-cell patients.

There is a need for biomarkers to predict and measure the severity of immune effector cell-associated neurotoxicity syndrome (ICANS). Glial fibrillary acidic protein (GFAP) and neurofilament light chain (NfL) are well-validated biomarkers of astroglial and neuronal injury, respectively. We hypothesized that pretreatment GFAP and NfL levels can predict the risk of subsequent ICANS and that increases in GFAP and NfL levels during treatment reflect ICANS severity. We measured cerebrospinal fluid GFAP (cGFAP) and NfL (cNfL) along with serum NfL (sNfL) levels at pretreatment and day 7 to 10 after chimeric antigen receptor (CAR) T-cell infusion in 3 pediatric cohorts treated with CD19- or CD19/CD22-directed CAR T cells. cGFAP and cNfL levels increased during grade ≥1 ICANS in patients treated with CD19-directed CAR T cells but not in those who received CD19/CD22-directed CAR T cells. The sNfL levels did not increase during ICANS. Prelymphodepletion cGFAP, cNfL, and sNfL levels were not predictive of subsequent ICANS. Elevated baseline cGFAP levels were associated with a history of transplantation. Patients with prior central nervous system (CNS) radiation had higher cNfL levels, and elevated baseline sNfL levels were associated with a history of peripheral neuropathy. Thus, cGFAP and cNfL may be useful biomarkers for measuring the severity of CNS injury during ICANS in children. Elevated baseline levels of cGFAP, cNfL, and sNfL likely reflect the cumulative injury to the central and peripheral nervous systems from prior treatment. However, levels of any of the 3 biomarkers before CAR T-cell infusion did not predict the risk of ICANS.

Comparison of Inhibitor and Substrate Selectivity between Rodent and Human Vascular Adhesion Protein-1.

Vascular adhesion protein-1 (VAP-1) is an ectoenzyme that functions as a copper-containing amine oxidase and is involved in leukocyte adhesion at sites of inflammation. Inhibition of VAP-1 oxidative deamination has become an attractive target for anti-inflammatory therapy with demonstrated efficacy in rodent models of inflammation. A previous comparison of purified recombinant VAP-1 from mouse, rat, monkey, and human gene sequences predicted that rodent VAP-1 would have higher affinity for smaller hydrophilic substrates/inhibitors because of its narrower and more hydrophilic active site channel. An optimized in vitro oxidative deamination fluorescence assay with benzylamine (BA) was used to compare inhibition of five known inhibitors in recombinant mouse, rat, and human VAP-1. Human VAP-1 was more sensitive compared to rat or mouse VAP-1 (lowest IC50 concentration) to semicarbazide but was least sensitive to hydralazine and LJP-1207. Hydralazine had a lower IC50 in rats compared to humans, although not significant. However, the IC50 of hydralazine was significantly higher in the rat compared to mouse VAP-1. The larger hydrophobic compounds from Astellas (compound 35c) and Boehringer Ingelheim (PXS-4728A) were hypothesized to have higher binding affinity for human VAP-1 compared to rodent VAP-1 since the channel in human VAP-1 is larger and more hydrophobic than that in rodent VAP-1. Although the sensitivity of these two inhibitors was the lowest in the mouse enzyme, we found no significant differences between mouse, rat, and human VAP-1. Michaelis-Menten kinetics of the small primary amines phenylethylamine and tyramine were also compared to the common marker substrate BA demonstrating that BA had the highest affinity among the substrates. Rat VAP-1 had the highest affinity for all three substrates and mouse VAP-1 had intermediate affinity for BA and phenylethylamine, but tyramine was not a substrate for mouse VAP-1 under these assay conditions. These results suggest that comparing oxidative deamination in mouse and rat VAP-1 may be important if using these species for preclinical efficacy models.