Patient Susceptibilities in Preclinical Drug Safety Assessment B. R. Berridge GlaxoSmithKline ILAR Roundtable October, 2017
Animal studies represent an apical preclinical modeling system in drug development Capabilities Animal studies Bioinformatics Phenotypic assays Activity assays Human tissue Target ID & validation Binding assays Target validation, PK, PK/PD, Efficacy, Safety Hit/lead discovery omics Lead optimisation Candidate selection #compounds 1000 s 100 s 10 s 1 3 Patient studies Preclinical safety Clinical assessment Target based drug development Evidence building Targets that modulate disease + Compounds that bind targets + Compounds that are active at the target + Compounds that are bioavailable + Compounds that are safe
Comparative mammalian anatomy and physiology is well conserved CV System as exemplar Blood vessels conduct blood to the heart itself as well as the rest of the body. Cardiomyocytes are contractile cells with immense energy needs Rhythmic waves of electrical activity ensure coordinated contraction of different regions of the heart. A muscular pump and its delivery system. Heart valves ensure unidirectional flow of blood.
Our models have been pretty good for acute liabilities
Rarely, severe toxicities do occur in Phase I trials
More concerning are chronic liabilities that manifest in susceptible patients.
Nat Rev Drug Disc 14: 475, 2015 Over half of development attrition is related to safety liabilities that occur either in animal studies or clinical trials
Differences in demographics Nonclinical Patients vs. Clinical Patients
Differences in clinical assessments Nonclinical Patients vs. Clinical Patients
Patients are complicated! Attributes of a diabetic patient population Concurrent medications Ischemic heart dz Stroke Heart failure Renal disease Hyperglycemia Hypertension Dyslipidemia Baseline characteristics of patients with diabetes studied in the RECORD trial. Lancet 373:2125 2135, 2009
But, animal models also have co morbidities Valvulopathy in rats and dogs Rodent progressive cardiomyopathy Chronic progressive nephropathy in rodents Vascular injury in rats and dogs
Patients are not all equally susceptible!
Improving our predictivity for liabilities in susceptible patients Potential approaches Maintain status quo Increasing the sensitivity of our assessments Animal models of human disease Stressed models Mechanistic approaches Non-animal or humanized systems
If we presume that the goal of animal testing is to provide an assurance of safety prior to phase I, that goal is supported by the high negative predictive value of animal safety studies. The high negative predictive value of animal safety studies satisfies that goal. The positive predictive value is undermined by a low prevalence as a result of not progressing molecules with significant safety liabilities. Given the way safety studies are designed, predictive values likely decline with clinical progression.
Next generation vivaria will enable collection of objective high content in vivo data. Body weight Body temperature Respiratory rate Activity HR, BP, ECG
Alternative models reveal new things In contrast, chronic treatment with a COX2 selective dose of rofecoxib exaggerated the harmful effects of the SFD, i.e. increasing vascular and renal dysfunction, dyslipidemia, hypertension Thus, MRI unmasked diastolic dysfunction during dobutamine stress. Dobutamine stress MRI allows noninvasive assessment of systolic and diastolic components of heart failure.
Hypertensive rats are more sensitive to doxorubicin cardiotoxicity.but, not to hematotoxicity!
Our attitudes toward alternative models
New technical capabilities Cisapride Bringing the human condition to bear on liability identification hipsc CMs from patients with heritable LQT, HCM, and DCM Pre existing disease induces susceptibility to arrhythmia
Mechanisms CV failure modes- Mechanisms to phenotypes Drug actions on human receptors, ion channels, cellular processes βar, PDE Na +, K + Ca 2+ ATP generation 5HT2B Cytotoxicity Etc. Potency + Exposure (dose, time) 1 o Failure modes Vasoactivity Action potential Inotropy Cardiomyocyte/ myocardial injury Valvular injury/proliferation Endothelial injury/coagulation Nonclinical Phenotypes BP Arrhythmia EF Myocardial necrosis Cardiac fibrosis Regurgitant flow Hemorrhage, thrombosis Clinical Phenotypes BP, HR, ctn, EF, HF, Arrhythmia, MACE
Mechanisms CV failure modes- Mechanisms to phenotypes Drug actions on human receptors, ion channels, cellular processes βar, PDE Na +, K + Ca 2+ ATP generation 5HT2B Cytotoxicity Etc. Potency + Exposure (dose, time) 1 o Failure modes Vasoactivity Inotropy Aim = shift our testing Action from phenotypic potential to mechanistic Cardiomyocyte/ myocardial injury Valvular injury/proliferation Endothelial injury/coagulation Nonclinical Phenotypes BP Arrhythmia EF Myocardial necrosis Cardiac fibrosis Regurgitant flow Hemorrhage, thrombosis Clinical Phenotypes BP, HR, ctn, EF, HF, Arrhythmia, MACE
Summary Our current approaches to preclinical drug safety assessment aren t designed to identify rare clinical events associated with individual patient susceptibilities. The evolution of precision medicine may change expectations. Advances in technical capabilities should change expectations. A component of progress in this area will require us to understand what constitutes susceptibility in a patient i.e. you can t model what you don t know.