pk/pd and safety assessment of biologics to support early clinical
TRANSCRIPT
PK/PD and safety assessment of biologics to support early clinical development
Jennifer Sims
Integrated Biologix GmbH, Basel, Switzerland AGAH Workshop, Munich, November 28-29, 2014
Critical Aspects of Integrated Drug Development – Expect the Unexpected!
Outline of presentation
• Mechanisms of toxicity • Predicting “on-target” clinical effects from
preclinical data • “Off target” binding and adverse effects • Immunogenicity related adverse effects • Integration of in vitro and in vivo exposure-
response relationships and NOAEL/HED/safety factor approaches to calculate safe starting dose for FIH study
Antibody derived scaffolds (protein engineering)
Fc-fusion proteins PEGylated proteins mAb Drug Conjugates ... ... ...
Mechanisms of Toxicity 1. Mechanism of action-related “On-target toxicity” 2. “Off-target binding / toxicity”
– Rare for biologics – but is frequency increasing? – Generally engineered out of candidate when discovered
3. Adverse consequences of immunogenicity – Translatability of ADA-related adverse effects for
humans?
4. Chemical-based toxicity of drug and/or metabolites – e.g. for Antibody Drug Conjugates
Mechanisms of Toxicity (1): “On-target toxicity”
Some examples • Immunostimulation: cytokine release (e.g. TGN1412), autoimmunity (e.g.
Graves Disease with alemtuzumab) • Immunosuppression: infections, malignancies • Cardiotoxicity: e.g. trastuzumab • Vascular neoplasms and liver/haematological changes with anti-DDL4 mAb • Developmental toxicity: e.g. oligohydramnios with trastuzumab &
pertuzumab
Hansel et al, The safety and side effects of monoclonal antibodies, Nature reviews 9, 325-338, 2010 Giezen et al, Safety-related regulatory actions for biologics approved in the US and EU, JAMA 300 (16): 1887-1896, 2008
Predicting “On-Target” Clinical Effects from Preclinical Data
6
Dose or Exposure
10 100 1000 10000
Effect
0
20
40
60
80
100
Therapeutic range
unacceptable toxicity
Toxic range
MTD NOAEL MABEL/PAD
Min Effective Dose (MED)
Acceptable effect
Understand shape / steepness of concentration-response relationships and MABEL/PAD
Translational PD biomarkers: - target engagement (eg receptor occupancy / ligand binding) - mechanism (eg downstream signalling) - outcome (eg pharmacological / clinical response) Use of a pharmacologically
responsive animal species Optimal study designs – use of all in
vivo and in vitro data
PK and total ligand profiles for a mAb against a soluble ligand (IL-1β)
Time (days)
0 20 40 60 80 100 120 140
Tota
l IL-
1 be
ta (p
M)
0.0
0.5
1.0
1.5
2.0
1.0 mg/kg 3.0 mg/kg10.0 mg/kg
• long half-life of mAb - long duration of effect
• slow elimination of mAb-ligand complex (generally much slower than ligand)
• saturable accumulation of total ligand – used as biomarker for target engagement
Nonlinear PK and % target saturation data for cynomolgus monkeys using a simple binding model-
• Non-linear PK in cynomolgus monkeys for mAb vs cell surface target - TMDD • Can be used to generate target saturation curves based on PK-PD model • Compare PK-PD model predictions with measured receptor occupancy if possible
Lowe et al 2009, Basic & Clinical Pharmacology & Toxicology 106, 195-209
Prediction of % target saturation from PK: Saturation (lines) was predicted form the nonlinear PK for cynomolgus
monkeys and patients, then overlaid with data (symbols) from an independent saturation assay on lymphocytes drawn from each individual
TGN1412: PK-target engagement data was not considered for clinical dose setting
• Cytokine storm observed in 6 healthy volunteers • Receptor occupancy assay showed evidence for CD28 target
saturation at all doses during toxicity study in cynomolgus monkeys
• No evidence for adverse effects in cynomolgus monkey one-month toxicity study Monoclonal antibody TGN1412 trial failure explained by species differences in CD28 expression on CD4+ effector memory T-cells
Eastwood et al, 2010. British Journal of Pharmacology 161: 512-526 • Target engagement assay (CD28 RO) used in 4-week toxicity study in
cynomolgus monkeys, but no in vivo pharmacological function biomarker and no comparative in vitro potency/functionality assay
• The cynomolgus monkey was not a pharmacologically relevant species
New clinical trial with TGN1412 Human regulatory T cells are selectively activated by low dose application of the CD28
superagonist TGN1412/TAB08, Tabares et al, 2014, Eur J Immunol
• Low doses (0.1 μg/kg to 7 μg/kg) of TAB08 given over a 4 to 12h IV infusion to healthy volunteers
• Starting dose 1000-fold less than 0.1 mg/kg in 2006 trial
• Dose-dependent systemic release of Treg-cell signature cytokine IL-10 in absence of pro-inflammatory factors associated with the CRS of the 2006 TGN1412 study
Selection of pharmacologically relevant species for translational PK-PD and safety assessment
High target selectivity
High species specificity
Human target ≠ Animal target
Predicting Clinical Effects from Preclinical Data
Protein Sequence Differences between Species / Subspecies
• Target sequence homology and polymorphisms across species and
strains • Cell / tissue expression of target, target turnover • Target binding affinity across species
Molecules do not act if they do not bind… but binding does not mean they act!
Different? Different? Pharmacologically relevant animal species
Predicting Clinical Effects from Preclinical Data: Determinants of species relevance
• Consider other species differences e.g.
• Fc effector function – FcRn, FcγRs, C1q binding, ADCC, CDC • Comparative biology, signaling pathways and downstream pharmacology
• In vitro functional assays for evaluating functional equivalence/difference across species
• In vivo PD markers / physiological / pharmacological outcome measure • Define safety assessment strategy
• rodent+non-rodent, non-rodent only (NHPs), alternative approaches – surrogate product, KI/KO transgenic mouse model, animal model of disease, in vitro approaches ….
Different? Different? Pharmacologically relevant animal species
Predicting Clinical Effects from Preclinical Data: Determinants of species relevance
Importance of appropriate in vitro data - examples
• Target binding characteristics / affinity / epitope • Comparative target expression (literature data plus RNA expression,
ISH, IHC) • Tissue cross reactivity / potential for unexpected binding • Relevant functional assays to understand relative species
differences and comparative biology • Receptor occupancy in human systems • FcγR binding and ADCC/CDC/apoptosis assays • Cytokine release in human systems • Special in vitro studies to address species differences in target
expression – E.g. platelet binding, platelet activation & aggregation studies – Binding to red blood cells
• “Predictive immunogenicity” – in silico, in vitro
Z. Waibler, L.Y. Sender, C. Kamp, J. Müller-Berghaus, B. Liedert, C.K. Schneider, J. Löwer, U. Kalinke, Toward experimental assessment of receptor occupancy: TGN1412 revisited, J. Allergy Clin. Immunol. 122 (2008) 890-892.
• Using human blood samples, the median of TGN1412–Alexa 488 staining gated on CD3+ cells was determined, and the percentage of specific binding was calculated
• Red bar represents range of TGN1412 concentrations theoretically present in the blood of a 70-kg human at a dose of 0.1 mg/kg TGN1412
• Receptor occupancy between 45% and 80% was obtained (compared to 90% estimated from static ligand binding model)
In vitro receptor occupancy
Assessing potential for a cytokine storm Hazard detection and characterization prior to clinical studies
• Experience has demonstrated that in vivo animal studies (even in NHPs) may not be good predictors of a cytokine storm in humans
• Many companies have developed in vitro cytokine release assays based on human blood or PBMCs to identify and characterize this hazard when appropriate based on target
Bugelski et al, 2010, Expert Rev Clin Immunol, 5: 499-521 Wolf et al, 2012 Cytokine, 60(3):828-37 Bailey et al, 2013 J Pharmacol Toxicol Methods. 68(2):231-9 Findlay et al, 2010. J Immunol Methods; 352: 1-12 Romer et al, 2011. Blood, 118 (26) 6772-82 Stebbings et al, 2013. J of Immunotox, 11(1):75-82 Stebbings et al, 2013. Brit J Clin Pharmacol 76(2) 299-315 Finco et al, 2014, Cytokine 66(2), 143-155.
Safety assessment of a T cell-engaging BiTE: bispecific binding to CD3 and EGFR
Lutterbuese et al, PNAS, 107 (28), 12605-12610
• Studies in cynomolgus monkeys – EGFR-BiTE administered to cyno monkeys by continuous i.v. infusion over 3-weeks – Groups of two monkeys received vehicle control or 6.2, 12.4, 31, or 154 μg/kg/day – At the higher doses of 31 and 154 μg/kg/d, severe signs of toxicity were observed
within 56h of start of dosing – Histopathological analysis showed signs of liver and kidney toxicity at 31 or 154 154
μg/kg/d - may be a result of redirected lysis of cells expressing low levels of EGFR in these organs
– Animals in both high-dose groups also showed increased levels of inflammatory cytokines in serum (i.e., TNF-α, IFN-γ, IL-6, IL-5, and IL-2)
– Histopathological changes including lymphocyte infiltration and cell death noted in all tissues known to express EGFR
• i.e., salivary glands, liver, stomach, small intestine, colon, rectum, kidneys, adrenal glands, ureter, urinary bladder, prostate, and epididymides.
– The relevance of the cyno for safety assessment and interpretation of histopathological changes is dependent on the comparability to humans of target binding (affinity, epitope, specificity), target expression and functionality
• The pattern of organ changes correlates with EGFR target distribution
Safety assessment of a T cell-engaging BiTE antibody: bispecific binding to CD3 and EGFR
Lutterbuese et al, PNAS, 107 (28), 12605-12610
Mechanisms of toxicity (2): Off target binding, unexpected PK and toxicity
• Biologics generally demonstrate high specificity for target
• Off target toxicity due to binding to unrelated targets not common for biologics
• Small but, increasing number of case studies of “unusual PK behavior” and “off-target” binding and adverse effects – Greater engineering of therapeutic protein
candidates?
Unexpected fast clearance of antibodies Hotzel et al, A strategy for risk mitigation of antibodies with fast clearance, mAbs 4:6,
753-760, 2012
• Pharmacokinetic data in cynomolgus monkeys collected for a panel of 52 Genentech mAbs showed broad distribution of target-independent clearance values (2.4–61.3 mL/day/kg) • 15 (29%) having clearance > 10 mL/day/kg
• Several palivizumab variants generated that enhanced the neutralization of RSV in vitro by up to 44-fold
• Unexpectedly, only a small increase of in vivo potency over palivizumab, - poor serum PK and lung bioavailability
• Unexpected broad tissue binding demonstrated • Changes at three amino acids arising from affinity
maturation markedly increased non-specific binding to various tissues
• Reversion of these three residues to the original sequences greatly diminished the tissue binding
Development of Motavizumab, an Ultra-potent Antibody for the Prevention of Respiratory Syncytial Virus Infection
in the Upper and Lower Respiratory Tract H Wu et al, J. Mol. Biol. 368, 652–665, 2007
Development of Motavizumab, an Ultra-potent Antibody for the Prevention of Respiratory Syncytial Virus Infection
in the Upper and Lower Respiratory Tract H Wu et al, J. Mol. Biol. 368, 652–665, 2007
Off target binding observed in human TCR study with mAb X
• Diffuse staining observed in keratinocytes of stratum corneum, granulosum and spinosum of the skin and Hassall’s corpuscles in thymus
• Further studies showed off target binding to a cytokeratin protein • mAb development terminated and new candidate selected with no off target
binding
Off-Target Platelet Activation in Macaques Unique to a Therapeutic Monoclonal Antibody Santostefano et al, Toxicol Pathol 40: 899, 2012
• AMG X: mAb vs soluble human
protein • Thrombocytopenia, platelet
activation, reduced mean arterial pressure, and transient loss of consciousness in cynomolgus monkeys after first IV dose
• In vitro, AMG X induced activation of platelets from macaque species but not from humans or baboons.
• Target protein not expressed on platelets
• Other similar mAbs against the same target failed to induce these in vivo and in vitro effects
Off target binding to plasma protein in humans and cynomolgus monkeys
• IgG1 mAb • No notable tissue binding in tissue cross reactivity TCR study with
cyno or human tissues • No AEs in cyno PK study (low dose) • 13-week repeated dose toxicity study in cynos
– Acute fatal reaction at high dose after first dose and thrombocytopenia and vascular changes at lower doses
– Off target binding identified by immunoprecipitation, SDS PAGE and LCMS – plasma protein, platelet factor-4
– Binding to PF-4 observed for cynos and humans but not rodents or dogs
– PF-4 cross reactivity leads to symptoms similar to heparin induced thrombocytopenia (anti-heparin-PF-4 complex autoantibodies)
• PF-4 cross reactivity successfully engineered out of follow-on candidate
Mitigation: screening for off target binding
• Assessment of binding to homologous targets and impact of plasma/serum on target binding affinity/function
• Protein chips – proprietary, commercial e.g. Protagen • ELISA-based “polyreactivity” screen • Assay based on ELISA detection of non-specific binding to baculovirus
particles (Hotzel et al, 2012 mAbs 4(6) 753-760). • Retrogenix platform
– arrays of expression vectors encoding more than 3500 human plasma membrane proteins – each membrane protein individually over-expressed in the context of human cells.
• Early exploratory tissue cross reactivity studies • TCR studies in panel of human (and animal) tissues post candidate
selection • In vivo PK behavior e.g. rodent PK screen for typical PK behavior • Risk mitigation: generally addressed by re-engineering candidate to
remove amino acid residues contributing to off target binding
Mechanisms of toxicity (3): Secondary to immunogenicity / anti-drug-antibodies
Consequences for safety: FDA draft guidance on immunogenicity, 2013 • Anaphylaxis • Cytokine release syndrome • “Infusion reactions” • Non-acute reactions
– Delayed hypersensitivity – Immune complex related toxicities
• Cross reactivity to endogenous proteins
“Studies in animals are generally limited in their ability to predict the incidence of human immune responses to a therapeutic protein, but they may be useful in describing the consequences of antibody responses, particularly when an evolutionarily conserved, non-redundant endogenous protein is inhibited by cross-reactive antibodies generated to its therapeutic protein product counterpart. “
Incidence of immunogenicity DRUG TARGET ADA in animals ADA in humans
Simulect CD25 75% in monkeys 0.3-3.5%
Campath CD52 100% in monkeys, rats and rabbits
1.4-65%, depending on indication
Tysabri VLA-4 32-75% in monkeys 6-10%
Orencia CD80/CD86 50% in mice, rats and monkeys during recovery phase
0.4-1.5%
Herceptin Her2 1.2% in monkeys 0.1%
Intron-A (IFNα)
Neutralising ADA in most animals
0-13%
Rebif (IFNβ)
Neutralising ADA in most animals
23-31%
Fabrazyme (α-galactosidase)
ADA in most rats in repeat dose studies
>80%
Aspariginase (E.coli)
Asparagine >70% in ALL patients
Consequences of immunogenicity: Translatability from nonclinical studies
• Cross reactivity with endogenous counterpart protein – Pegylated MDGF
• ADA and nAbs versus MDGF in mice and monkeys • Cross reactive nAbs to endogenous thrombopoietin and severe thrombocytopenia • Low incidence of nAbs in humans leading to severe thrombocytopenia
– Recombinant erythropoietin and Factor VIII products • Neutralizing cross reactive antibodies developed in animal studies
• Cross-linking of target receptor – VH domain antibody TNFR antagonist – cytokine release in vitro and in healthy
volunteers • Hypersensitivity and immune complex pathologies observed in animal
studies with biologics: – Leach et al, 2014 Toxicologic Pathology 42:293 – Jojko et al Toxicol Pathol, published online 3 April 2014
Translatability of immune complex related findings for drugs showing ADA formation in patients?
Cytokine release in healthy volunteers with novel scaffold: anti-TNFR1 VH domain antibody (GSK1995057)
Holland et al. J Clin Immunol, July 2013
• About 50 % human subjects had pre-existing human anti-VH antibodies • Formation of HAVH autoantibody/GSK1995057 complexes activated TNFR1
and caused cytokine release in vitro in some, but not all, of the human cell types tested
• No evidence of TNFR1 activation or agonism was observed in preclinical studies of GSK1995057 in Cynomolgus monkeys.
• Healthy males subjects received a single GSK1995057 intravenous infusion of 0.0004, 0.002 and 0.01 mg/kg.
• All enrolled subjects were pre-screened for human anti-VH (HAVH) autoantibody status and prospectively stratified accordingly.
• Starting dose selected to achieve GSK1995057 plasma concentrations below the threshold required to stimulate TNFR1 agonism in vitro in the presence of HAVH autoantibodies
• Higher provisional dose levels were specified in the original protocol but were not reached due to dose-limiting toxicity observed in a subject in Cohort 3b (0.01 mg/kg)
• Study design reflected the potential risk of cytokine release in HAVH-positive subjects – incorporated assessment of GSK1995057 tolerability in five subjects who were
negative for pre-existing HAVH autoantibodies prior to dosing and five subjects who tested positive in each dosing cohort
Cytokine release in healthy volunteers with novel scaffold: anti-TNFR1 VH domain antibody (GSK1995057)
Holland et al. J Clin Immunol, July 2013
• Clinical and physiological signs of cytokine release were observed in two HAVH autoantibody-positive healthy subjects following GSK1995057 infusion
• In vitro, HAVH autoantibody levels correlated with TNFR1- dependent cytokine release and propensity for cytokine release in humans following GSK1995057 dosing.
Ab1: 0.002 mg/kg Ab2: 0.01 mg/kg
Hypersensitivity reactions (HSRs) and immune complex-related pathologies
Jojko et al Toxicol Pathol, published online 3 April 2014
• Case studies presented of IC-related pathology in monkeys and rodents • Post dose reactions / mortality after +3 doses, with no relation to dose
level • Vascular inflammation, glomerulonephropathies • Histopathology and IHC suggests these effects may be mediated by
deposition of immune complexes (ICs) • ICs may be observed in glomerulus, blood vessels, synovium, lung, liver,
skin, eye, choroid plexus etc, or bound to neutrophils, monocytes/macrophage, or platelets
• IC deposition may activate complement, kinin and or coagulation/fibrinolytic pathways and result in a systemic pro-inflammatory response
• Potential risk of IC formation - translatability to human risk in clinical trials? – Generally thought that immunogenicity observed in preclinical species might
not be considered predictive to humans (for human and humanized proteins – ICH S6 R(1))
Hepatic necrosis in a repeated dose rat toxicity study with a mAb versus cell membrane receptor
• Finding observed only in low dose group and only by IV route, not SC route • Hepatic effects occurred after 3-4 weekly doses, shortly after dosing (higher ALT/AST
present at about 6h post dose but not at 5 days later) • High incidence and titer of ADA by both IV and SC • Areas of necrosis positive for drug and rat IgG, IgM and C3 by IHC
Drug-ADA-associated kidney findings in a cynomolgus monkey
Vascular/perivascular inflammation in kidney (H&E stain) – fibrinoid/hyaline change (asterisks) in a small artery/large arteriole with mononuclear cells in outer media and adventitia
IHC staining for drug reveals granular deposits containing drug and located in intima and media.
Effect of IC lattice formation on IC tissue deposition and pathogenicity
Integrated weight of evidence approach to evaluate HSRs / IC-related adverse effects
• Use weight of evidence to support adverse effects being secondary to ADA and IC formation – Exclude MOA-related mechanism – Temporal relationship of clinical sings to dosing – Presence of ADA and/or circulating immune complex assays – Impact on PK - accelerated clearance of drug – Complement activation (CH50, complement split products) – Immunohistochemistry for drug, IgM, IgG, sC5b-9, C3a – Presence of drug-containing granular deposits in proximity to
H&E lesions – often associated with intima and/or media of small vessels
– TEM for renal glomerular complexes • Impact on NOAEL and starting dose for FIH study? • Risk mitigation strategy for clinical studies
Summary: integrated pharmacological-toxicological assessment
In vitro animal
In vitro human
Pharmacology / toxicology
Species selection
In vivo toxicology studies
Safety assessment Efficacy assessment
Adapted from Benno Rattel
Target-related risks? Target expression - RNA expression - IHC / TCR
Human tissue panel
Animal tissue panel
In vivo Ex vivo
PK and PD assays ADA assays
Proof of exposure / PK-PD relationships
PK-PD relationships
Pharmacology
[5] PAD / MABEL
Summary: Justification of starting dose in humans
Toxicology - Relevant species - Dosing regimen based on PK / duration
of effect [1] NOAEL (or HNSTD for oncology) [2/3] HED - adjust for anticipated exposure in man?
[4] Apply >10-fold safety factor
“Maximum Recommended Starting Dose”
* NB an additional factor may be added based on uncertainty of data / prediction and relative risk
- Justify based on pharmacology (e.g. minimally / optimally effective dose)
- Receptor occupancy - In vitro data in human systems - Adjust for anticipated exposure in man - Account for anticipated duration of
effect - Adjust for inter-species differences in
affinity / potency
Pharmacology
[5] PAD / MABEL
Toxicology - relevant species - dosing regimen based on PK / duration
of effect [1] NOAEL (or HNSTD for oncology) [2/3] HED - adjust for anticipated exposure in man?
[4] Apply >10-fold safety factor
“Maximum Recommended Starting Dose”
* NB an additional factor may be added based on uncertainty of data / prediction and relative risk
Translatability of ADA-related toxicity and impact on NOAEL?
- Justify based on pharmacology (e.g. minimally or optimally effective dose)
- Receptor occupancy - In vitro data in human systems - Adjust for anticipated exposure in man - Account for anticipated duration of
effect - Adjust for inter-species differences in
affinity / potency
Summary: Justification of starting dose in humans
