Abstract 3694: Clinical validation of a quantitative systems pharmacology (QSP) model of ISB 2001 used for deriving first in human (FIH) dose and efficient phase 1 dose escalation design in relapsed refractory multiple myeloma (RRMM) patients
Abstract Background: ISB 2001 is the first trispecific antibody that simultaneously targets BCMA and CD38 on MM cells, and CD3 to engage T-cells, forming tumor/immune cell synapses (TICS) that cause tumor cytotoxicity. Since ISB 2001 lacks cross-reactivity to common preclinical toxicology species including monkeys, no preclinical toxicology studies were conducted to inform the FIH dose calculation. The conventional minimal anticipated biological effect level (MABEL) approach for FIH dose derivation for T-cell engagers (TCE) often results in exposing patients to non-therapeutic doses. Instead, we developed a QSP model to predict an optimal FIH dose and efficacious dose range. This informed the design of the ongoing phase 1 dose escalation study of ISB 2001 in RRMM patients (NCT05862012), starting at 5 µg/kg subcutaneously with the initial three single-patient cohorts followed by standard 3+3 cohorts beginning at 150 µg/kg. Method: A QSP model utilizing binding affinities, target expressions, cell counts, and preclinical pharmacokinetic (PK) data was developed to characterize TICS profiles in the tumor compartment. Normalized TICS formation was linked to tumor cytotoxicity, and the model was calibrated with ISB 2001 in vivo efficacy data to enable efficacy predictions. This QSP model was benchmarked with teclistamab’s clinical data to derive the minimal pharmacologically active dose (MPAD) of ISB 2001[1]. The MPAD was further adjusted using in vitro cytokine secretion data to establish the FIH dose, which was preceded by fractionated step-up doses to mitigate potential cytokine release syndrome concerns. Serum concentration of ISB 2001 across doses up to 1200 µg/kg were quantified using a validated electrochemiluminescence assay and compared with the model predictions. Results: No CRS was observed in the FIH cohort, and only one grade 1 case was reported at doses up to 15 µg/kg in third cohort, supporting the QSP model-based FIH dose selection while acknowledging the limited sample size. As predicted, objective tumor responses were observed at dose levels as low as 50 µg/kg, just 10-fold higher than the FIH dose. Measured ISB 2001 serum concentration time profiles across multiple dose levels up to 1200 µg/kg (n=20) aligned well with model simulations. Conclusion: Despite the absence of monkey PK and toxicology data, this QSP-guided approach successfully predicted an optimal FIH dose, clinical PK, safety, and efficacy, while minimizing patient exposure to non-therapeutic doses. With this method, TCEs can be optimized solely for activity against human tumors without requiring cross-reactivity to animal species. Overall, the QSP based FIH calculation approach offers an effective pathway for designing phase 1 studies for future TCEs.
Read more