Predicting Optimal Drug Parameters and Doses


The goal of this collaboration was to provide early quantitative decision-making guidance for the project team by developing and interrogating a quantitative systems pharmacology (QSP) model of the co-modulation inhibitory receptors PD-1 and TIM-3 in immuno-oncology.

The QSP model was to explain why marketed anti-PD1s have such similar dose regimens despite very different Kds (MK-3475 (Pembrolizumab, Keytruda™): 2mg/kg Q3W, IV, Kd = 30pM and MDX-1106 (Nivolumab, Opdivo™): 3mg/kg Q2W or Q3W, IV, Kd = 3 nM) as well as predict optimal drug properties targeting PD-1 and TIM-3 in oncology for bispecific biologics as well as fixed-dose combinations (FDC). We also provided a bispecific versus FDC risk assessment.


  1. QSP model analysis predicted there would be diminishing returns on very tight binding biologics due to Target Mediated Drug Disposition (TMDD) that offsets potency


  2. QSP model analysis predicted there is no advantage between FDC, 2-2 bispecific, and 2-1 bispecific formats, which are predicted to be roughly equivalent in terms of dosing requirements

The Model

The QSP model was based on first principles as a system of elementary mass-action, mechanistic PKPD, ordinary differential equations. The model reactions include protein synthesis and elimination, ligand-receptor and drug-target formation and turnover, and drug administration and first order clearance. 


Tim3/PD1 Pharmacology Schematic



  • Four versions of the model were created: PD-1 monospecific, TIM-3 monospecific, PD-1 x TIM-3 bispecific, and FDC targeting PD-1 and TIM-3

  • Models were benchmarked against in vitro, in vivo, and clinical data.

  • Parameters were set to known physiological values and unknown parameters were estimated via parameter estimation.  

  • Model correctly predicted approved efficacious doses for MK-3475 (Pembro) and MDX-1106 (Nivo) and explained the counterintuitive result that 100x difference in affinity results in only a modest reduction in dose

  • The model was used to generate a list of sensitive parameters which helped prioritize which parameters were most important to understand and acquire data for, to prioritize experiments. 

  • The model was used to compare various FDC and bispecific platforms and identified optimal drug characteristics for dosing requirements


Exploring Model Avidity



  1. Generated hypothesis for dose equivalence for ant-PD-1s despite very different potencies (Kds)


  2. FDC, 2-2, 2-1 formats were roughly equivalent in terms of allowing best opportunity for low dosing


  3. Optimal affinities and avidities were identified for each platform


  4. Analysis revealed knowledge gaps and prioritized experiments


  5. Simulated PK study for Tim3 parameters



Download the Case Study