Predicting Tissue Distribution & Clearance of Antibody Formats to Improve Selectivity of Targeted Therapies

Transcription

1 Predicting Tissue Distribution & Clearance of Antibody Formats to Improve Selectivity of Targeted Therapies Ben-Fillippo Krippendorff, Roche Innovation Center Basel

2 Word cloud generated from ADC world survey: «What characteristics of ADCs are critical to facilitate their translation to the clinic? Data from

3 response (%) Is the dose-response curve the full story? Efficacy in the patient ED50 = 0.3 mg/kg ED50 = 3 mg/kg Dose (mg/kg)

4 response (%) Is the dose-response curve the full story? Efficacy and dose limiting toxicity ED50 = 0.3 mg/kg TD50 = 0.6 mg/kg ED50 = 3 mg/kg TD50 = 18 mg/kg Therapeutic window Dose (mg/kg) 1. A wide therapeutic window increases probability of success in a diverse patient population 2. Early predictions of this window facilitate good decision making

5 Local concentrations often drive toxicities in tissues

6 Are effect and toxicity of ADC linked to exposure? 1. Dose drives tissue concentrations 2. Cellular uptake in the tissues drives efficacy as well as toxicity. Early toxicity observed

7 Predicting relevant local concentrations by using Physiologically Based Pharmacokinetic Models (PBPK) Physiological parameters (drug-independent) Tissue volumes and blood flows Tissue composition Intestinal ph, transit times Enzyme and transporter abundance Populations Age, gender, race Disease status Genetic status Smoking/diet Drug-dependent parameters Small molecules Large molecules Differences of Patients and Preclinical species Models available for mice, rat, monkey, dog, minipig PBPK models are especially useful for species translation and coupling with effect and toxicity models

8 PBPK model for large molecules Can we better understand how the concentration at the target tissue is linked to desired and undesired effects? Physiological parameters (drug-independent) Drug-dependent parameters Populations and disease status Differences of Patients and Preclinical species Physiologically based pharmacokinetic and cellular uptake models Ferl et al. Ann Biomed Eng Fronton et al. J Pharmacokinet Pharmacodyn 2014 Krippendorff et al, J Pharmacokinet Pharmacodyn 2012 Krippendorff et al, J Pharmacokinet Pharmacodyn 2009

9 %ID/g %ID/g %ID/g %ID/g %ID/g %ID/g Predicting relevant concentrations by using Physiologically Based Pharmacokinetic Models (PBPK) Plasma 10 8 Liver 60 6 Physiological parameters (drug-independent) hours Spleen hours Kidney Drug-dependent parameters Populations and disease status hours Lung hours hours Bone hours Differences of Patients and Preclinical species But: How can we better understand the difference between Tissue concentrations vs cellular uptake and degradation?

10 Combining PBPK and Genentech data of a nonbinding antibody with two different labels I-125 label = where is the antibody in the moment In-111 label = where is the antibody in the moment + where was the antibody degraded in the past

11 In every tissue Large molecule PBPK modified from Ferl et al. with cellular uptake Blood circulation Vascular Interstitial Endosome free FcRn FcRn bound Internalized /degraded Vascular in the tissue = fixed from physiological data = fitted for each tissue separately = fitted one parameter for all tissues degraded Residualizing signal = + + Non - Residualizing signal = +

12 PBPK model predicted tissue toxicities for FcRn - (Prediction, no fitting) The FcRn KO variant is predicted to result in higher peak levels of antibody internalization in Liver and Spleen (the organs with high toxicity).

13 PBPK model for large molecules Normal tissue distribution of antibody fragments without target and influence of molecular size Influence of molecular size on tissue distribution of antibody fragments Zhe Li*, Ben-Fillippo Krippendorff*, Sharad Sharma, Antje C. Walz, Thierry Lavé, Dhaval K. Shah; mabs Journal 2015 In collaboration with

14 Tissue distribution PBPK model for large molecules Predicting tissue uptake of protein therapeutics based on size Most tissues: BC50 ~ 35 kda Size (kda)

15 Summary Physiological based pharmacokinetic model (PBPK) can help to predict tissue concentrations of ADC and therefore help to explain dose-limiting toxicity. For ADC it is important to remember that measured exposure might not correlate with toxicity as good as uptake and clearance in tissues. Biodistribution coefficients (BC) allow to estimate the tissue concentrations of ADC when only plasma concentration was measured. These biodistribution coefficients are even available for antibody fragments Outlook: We are working on linking these BC values for fragments with clearance data to predict which formats are optimal for targeting or avoiding different tissues.

16 Acknowledgements Buffalo University Prof. Dhaval Shah Zhe Li Sharad Sharma Potsdam University Prof. Wilhelm Huisinga Pharmaceutical Sciences, pred, Roche Innovation Center Basel Antje Walz Michael Gertz Neil Parrott Hans Peter Grimm Nicolas Frances Miro Eigenmann Genentech Greg Ferl Saroja Ramanujan Andrew Boswell Thierry Lavé Franz Schuler Michael Otteneder Lisa Benincosa Thomas Singer

17 Doing now what patients need next