Understanding and predicting dose dependent and idiosyncratic events
12-07-2016-1330-Paul-Watkins-YES
12-07-2016-1330-Paul-Watkins-YES
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
®<br />
Modeling Hepatoxicity:<br />
<strong>Underst<strong>and</strong>ing</strong> <strong>and</strong> <strong>predicting</strong> <strong>dose</strong><br />
<strong>dependent</strong> <strong>and</strong> <strong>idiosyncratic</strong> <strong>events</strong>.<br />
1
Disclosure<br />
I chair the scientific advisory committee for the<br />
DILIsim Initiative <strong>and</strong> own<br />
equity in the commercial spin off company<br />
DILIsym Services Inc.<br />
2
Outline of Talk<br />
1). Problem of liver safety<br />
2). The DILIsim Initiative<br />
3). Progress/challenges<br />
4). Conclusions
Reasons for Termination of Programs due to Safety by<br />
Organ System<br />
4
Posted: 5:17 p.m. Friday, Nov. 4, 2016<br />
The Associated Press<br />
WASHINGTON —<br />
Cempra is one of a h<strong>and</strong>ful of drugmakers developing new<br />
antibiotics amid growing bacterial resistance to decades-old<br />
drugs like penicillin.<br />
On Wednesday Cempra shares plunged more than 60<br />
percent after the FDA posted an online review highlighting<br />
irregular liver enzyme measurements reported with the drug,<br />
called solithromycin……..<br />
Copyright The Associated Press
Outline of Talk<br />
1). Problem of drug safety<br />
2). The DILIsim Initiative<br />
3). Progress<br />
4). Conclusions
UNC Institute for Drug Safety Sciences<br />
In Silico Modeling<br />
DILIsym ®<br />
Patients<br />
In Vitro<br />
Cutting Edge<br />
Pre-clinical Models<br />
7
DILIsym ® : 'Middle Out' <strong>and</strong> Multi-Scale<br />
Mitochondrial dysfunction<br />
Cellular life-cycle<br />
Drug distribution<br />
& metabolism<br />
Patient variability<br />
(SimPops)<br />
Kuepfer 2010, Molecular Systems Biology<br />
8
Approach<br />
1). Build mechanistic “modules” using differential equations<br />
– perform experiments to fill in knowledge gaps.<br />
2). Integrate the modules with the outcome of hepatocyte<br />
death <strong>and</strong> release of traditional <strong>and</strong> novel serum<br />
biomarkers.<br />
3). Vary model parameters to create simulated patient<br />
populations (Simpops®)<br />
4). Refine the aggregate model through incorporating data<br />
obtained from successive “exemplar” drugs
DILIsym ® Modules<br />
10
Outline of Talk<br />
1). Problem of drug safety<br />
2). The DILIsim Initiative<br />
3). Progress/challenges<br />
4). Conclusions
DILIsym ® Modules<br />
12
Drugs Can Inhibit Bile Acid Transporters<br />
Hepatotoxicity<br />
MRP3/4<br />
BSEP<br />
Bile<br />
Acids<br />
Drug<br />
OĀ<br />
NTCP<br />
(OATP)<br />
Bile<br />
Acids<br />
Drug
Drug PBPK model<br />
BA Transport Inhibition Model<br />
Bile Acids <strong>and</strong> DILI<br />
Drug inhibits BA<br />
transporters<br />
Bile Acid Homeostasis Model<br />
Cellular ATP Model<br />
Bile acid<br />
accumulation<br />
disrupts cellular<br />
energy balance<br />
BA H+<br />
Gradient<br />
Uncoupling<br />
Hepatocyte Life-Cycle<br />
Disrupted cellular<br />
energy balance can<br />
cause hepatocyte<br />
necrosis<br />
14
AMG 009<br />
No evidence of liver injury in multiple species<br />
• Rats, mice, non-human primates, hamsters or rabbits<br />
– During Phase I clinical trials in healthy volunteers, 5/8 patients<br />
showed significant <strong>and</strong> reversible transaminase elevations<br />
– Development of AMG 009 was halted<br />
– Bile acid transporter inhibition was the only mechanism<br />
identified as likely contributors to AMG 009 hepatotoxicity<br />
• No reactive metabolites, covalent binding, or mitochondrial toxicities<br />
were detected<br />
15
Some AMG 009 Transporter Inhibition Data<br />
X<br />
Bile acids<br />
AMG009<br />
Ryan Morgan - Amgen
Modeling in vitro transporter<br />
inhibition<br />
Bile acids<br />
Drug<br />
Transporter<br />
Inhibition<br />
constants<br />
Model average<br />
human<br />
Model simulated<br />
patient<br />
population<br />
(Simpops TM )
AMG 009 Modeling Approach<br />
PBPK Modeling<br />
• In vitro PK data<br />
• In vivo PK profile<br />
• Parameter optimization<br />
Mechanistic<br />
Toxicity Data<br />
•Transporter inhibition<br />
for AMG 009 (Ki, type of<br />
inhibition)<br />
PBPK model <strong>and</strong> toxicity data<br />
inputs were not modified after<br />
hepatotoxicity simulations<br />
Stop criteria<br />
used*<br />
Hepatotoxicity<br />
Simulation<br />
• Baseline<br />
• Population (SimPops)<br />
Combine PK <strong>and</strong> in vitro Tox<br />
Further optimization using<br />
AMG 009/853 clinical<br />
toxicity data was not<br />
necessary<br />
*<br />
ALT screening before AM dosing on days 2,3,4,5,8,11, <strong>and</strong> 14<br />
If ALT>3X ULN, dosing discontinued after 24 h<br />
Employed to recapitulate the clinical protocol of AMG 009<br />
Compare to clinical data<br />
Clinical Data <strong>and</strong><br />
Simulation Results<br />
18
DILIsym ® Predicts Dose-Dependent AMG 009 Hepatotoxicity in<br />
Human SimPops<br />
HUMANS<br />
100 mg BID 14 d Dose-Response<br />
(25-100 mg BID 14 d)<br />
Treatment<br />
Stop criteria<br />
used<br />
• DILIsym ® predicts <strong>dose</strong>-<strong>dependent</strong>, delayed<br />
presentation of AMG 009 hepatotoxicity <strong>and</strong><br />
recovery after discontinuation<br />
• Incidence rates were fairly similar to observations<br />
Clinical Data <strong>and</strong><br />
Simulation Results<br />
19
No Hepatotoxicity Predicted in the<br />
Rat SimPops Administered AMG 009<br />
1500 mg/kg/day PO for 1 month<br />
1500 mg/kg PO 1 month †<br />
Direct BA Toxicity Model<br />
No. Deaths 0/187<br />
ALT > 3X 0/187<br />
Bili > 2X 0/187<br />
Treatment<br />
RATS<br />
Preclinical Data <strong>and</strong><br />
Simulation Results<br />
20
AMG 853<br />
• AMG 853 was the backup to AMG 009 advanced into<br />
the clinic<br />
– No evidence of liver injury in preclinical species.<br />
– No evidence of human toxicity in Phase 1 or 2<br />
clinical trials<br />
– AMG 853 was a more potent BSEP inhibitor than<br />
AMG 009 with a Ki = 1.8 µM vs 2.4 µM<br />
21
DILSYM Modeling of AMG 853<br />
• DILIsym predicted that AMG 853 was safe in simulated<br />
humans (SimPops) although you would predict toxicity<br />
based on the BSEP Ki value…..<br />
Why? -<br />
Inhibition type was the key
Why mechanism of transport inhibition matters<br />
Hepatotoxicity<br />
MRP3/4<br />
BSEP<br />
Bile<br />
Acids<br />
Drug<br />
OĀ<br />
NTCP<br />
(OATP)<br />
Bile<br />
Acids<br />
Drug
Conclusion<br />
QSP modeling was able to predict the<br />
hepatoxic potential of AMG 009 <strong>and</strong> the safety<br />
of AMG853 based on in vitro<br />
liability assessments
Posted: 5:17 p.m. Friday, Nov. 4, 2016<br />
The Associated Press<br />
WASHINGTON —<br />
On Wednesday Cempra shares plunged more than 60<br />
percent after the FDA posted an online review highlighting<br />
irregular liver enzyme measurements reported with the drug,<br />
called solithromycin……..<br />
What exactly did the FDA say?
What the FDA Briefing Document Said<br />
“serious liver safety concern” <strong>and</strong> suggested two pathways forward:<br />
1). “the number of study subjects treated with solithromycin should be<br />
increased from 924 to approximately 12,000 <strong>and</strong> carefully assessed for<br />
liver safety <strong>events</strong>, either in exp<strong>and</strong>ed Phase III r<strong>and</strong>omized trials or in a<br />
large clinical safety study, in advance of making a regulatory decision<br />
regarding approval”<br />
2). “Alternatively, if solithromycin treatment has been found to convincingly<br />
offer a clinically substantial benefit over other currently approved<br />
treatments a second avenue might be considered….”
Two issues with solithromycin:<br />
– Imbalance in serum ALT elevations<br />
– Structural similarity with telithromycin
Structures of Macrolide Antibiotics
DILIsym ® Mechanism-Based Modeling<br />
Drug Properties<br />
• Oxidative stress<br />
• Mitochondrial dysfunction<br />
• Bile acid transporter inhibition<br />
Human population<br />
Liver Exposure<br />
PBPK Modeling<br />
• Compound Properties<br />
SimPops®<br />
• Tissue penetration studies<br />
• Pharmacokinetic data<br />
• in vitro data<br />
30
The rates of serum ALT elevations in clinical trials are<br />
reasonably predicted by DILIsym<br />
Compound<br />
Protocol<br />
Peak ALT > 3X ULN<br />
Observed<br />
Simulated<br />
Solithromycin<br />
Erythromycin<br />
Oral<br />
(CE01-300)<br />
IV-to-Oral<br />
(CE01-301)<br />
500 mg<br />
QID 10 days<br />
5.4% (3.2%) 3.9%<br />
9.1% (5.5%) 6.0%<br />
1-2% 2.8%<br />
Clarithromycin 500 mg BID 7 days 1-2% 2.8%<br />
Simulation Results <strong>and</strong><br />
Clinical Data<br />
Telithromycin 800 mg QD 10 days 0.0-0.8% 0%<br />
data<br />
Data presented at Nov 4 2017 anti-infective Ad com<br />
31
Contribution to Predicted ALT elevations<br />
in Simulated Human Population<br />
DILI Mechanism Solithromycin Telithromycin Erythromycin Clarithromycin<br />
Mitochondrial<br />
Respiration<br />
Inhibition<br />
Predominant None None Predominant<br />
Oxidative<br />
Stress<br />
None None Minor None<br />
Bile Acid<br />
Transporter<br />
Inhibition<br />
Minor (Predominant) Predominant Minor<br />
Data presented at Nov 4 2017 anti-infective Ad com<br />
32
Excerpt from FDA Briefing Document<br />
…. A somewhat surprising additional<br />
unexplained gap in the analysis submitted by<br />
DILIsym Services is the absence of the parallel<br />
testing of telithromycin hepatoxicity in a<br />
simulated CAP population……The use of<br />
telithromycin as a “positive control” in the<br />
model with comparative liver test data would be<br />
highly relevant <strong>and</strong> might support the utility of<br />
the model….”
Contribution to Predicted ALT elevations<br />
in Simulated Human Population<br />
DILI Mechanism Solithromycin Telithromycin Erythromycin Clarithromycin<br />
Mitochondrial<br />
Respiration<br />
Inhibition<br />
Predominant None None Predominant<br />
Oxidative<br />
Stress<br />
None None Minor None<br />
Bile Acid<br />
Transporter<br />
Inhibition<br />
Minor Predominant Predominant Minor<br />
Data presented at Nov 4 2017 anti-infective Ad com<br />
34
Summary<br />
– Imbalance in serum ALT elevations<br />
• Well-characterized clinically <strong>and</strong> mechanistically<br />
– Structural similarity with telithromycin<br />
• Measured solithromycin effects on the liver differ
FDA panel narrowly backs Cempra antibiotic<br />
Posted: 5:17 p.m. Friday, Nov. 4, 2016<br />
The Associated Press<br />
WASHINGTON —<br />
A panel of federal health advisers has narrowly recommended<br />
approval for an experimental antibiotic from Cempra Inc., a small<br />
North Carolina drugmaker.<br />
The Food <strong>and</strong> Drug Administration's outside experts voted<br />
7-6 in favor of the drug, saying its effectiveness outweighed<br />
risks of liver toxicity seen in company studies. The vote is<br />
nonbinding but the FDA often follows the advice of its panelists.<br />
Copyright The Associated Press
DILIsym ® Sub-models<br />
37
Multiple Steps Involved in<br />
Idiosyncratic DILI<br />
38
Multiple Steps Involved in<br />
Idiosyncratic DILI<br />
39
Multiple Steps Involved in<br />
Idiosyncratic DILI<br />
40
FDA Submission: A New Antibiotic<br />
that Caused Serum ALT Elevations<br />
in Phase 1 Study<br />
X X X<br />
Serum<br />
biomarkers<br />
ccK18<br />
CK18<br />
Oxidized<br />
HMGB1<br />
No acetylated<br />
HMGB1<br />
interpretation apoptosis<br />
41
Summary<br />
New mechanistic biomarkers promise to<br />
revolutionize the interpretation of liver safety<br />
in clinical trials <strong>and</strong> ultimately in the clinic.<br />
42
Outline of Talk<br />
1). Problem of drug safety<br />
2). The DILIsim Initiative<br />
3). Progress/challenges<br />
4). Conclusions
DILIsym ® Mechanism-Based Modeling<br />
Drug Properties<br />
• Oxidative stress<br />
• Mitochondrial dysfunction<br />
• Bile acid transporter inhibition<br />
Human population<br />
Liver Exposure<br />
PBPK Modeling<br />
• Compound Properties<br />
SimPops®<br />
• Tissue penetration studies<br />
• Pharmacokinetic data<br />
• in vitro data<br />
44
Optimal requirements for the cell system<br />
1). Metabolically competent<br />
2). Measure intracellular concentrations of<br />
parent <strong>and</strong> metabolites<br />
3). Identify the three major mechanisms<br />
4). Provide quantitative parameters for<br />
direct incorporation into DILIsym<br />
including PB/PK parameters<br />
5). Validation with the ~30 drugs successfully<br />
modeled already in DILIsym<br />
45
Confidence in Predictions<br />
Depends on Confidence in Data Used<br />
in vitro<br />
tox<br />
in vivo<br />
tox<br />
in vitro<br />
mechanistic<br />
in vivo<br />
mechanistic<br />
Clinical<br />
mechanistic<br />
Sims required for confidence in predictions<br />
Confidence in data for predictions
Outline of Talk<br />
1). Problem of drug safety<br />
2). The DILIsim Initiative<br />
3). Successes <strong>and</strong> Challenges<br />
4). Conclusions
Conclusion<br />
Quantitative Systems Pharmacology<br />
is already having an impact on drug<br />
development decisions both within<br />
industry <strong>and</strong> regulatory agencies.<br />
The impact should rapidly increase in<br />
the years ahead.
49<br />
The DILI-sim Team Sept 2015
Special thanks to the DILIsim Partners<br />
DILIsym ®<br />
And Cindy Ashfari <strong>and</strong> Ryan Morgan - Amgen<br />
50
Thanks<br />
UNC Institute for Drug Safety<br />
Sciences Leaders<br />
DILIsim Team <strong>and</strong> SAB<br />
Mouse Genetics: Merrie Mosedale<br />
Human Genetics: Tom Urban<br />
Organotypic liver cultures: Ed LeCluyse<br />
Cellular Imagine Core: Joe Trask<br />
UNC –Hamner Organ Toxicity<br />
Biomarker Core: Rachel Church<br />
UNC-Hamner Metabolomics Core- Jeff<br />
MacDonald<br />
UNC Chapel Hill<br />
Kim Brouwer<br />
U. of Liverpool<br />
Kevin Park<br />
Dean Naisbitt<br />
Daniel Antoine<br />
Munir Pirmohamed<br />
Qualyst<br />
Ken Brouwer<br />
51
Back up
DILIsym ® Performance Review – Level 2<br />
• Key Question: how do the frequency, <strong>dose</strong> response effect, <strong>and</strong> mechanisms predicted compare to the observed cases?<br />
Drug<br />
Injury<br />
Frequency<br />
Injury Dose-<br />
Response<br />
Injury<br />
Severity<br />
Injury Timing<br />
Injury<br />
Mechanism<br />
Entacapone (Clean)<br />
Tolcapone (DILI)<br />
Methapyrilene (Clean)<br />
Troglitazone (DILI)<br />
Pioglitazone (Clean)<br />
AMG009 (DILI)<br />
Compound A (DILI)<br />
Bosentan (DILI)<br />
Telmisartan (Clean)<br />
Compound D (DILI)<br />
Compound B (DILI)<br />
Color Key – Accuracy of DILIsym®<br />
Excellent<br />
Good<br />
Fair<br />
Poor<br />
Unavailable<br />
HUMAN<br />
Compound C (DILI)<br />
Etomoxir (DILI)<br />
Compound E (DILI)<br />
Clinical Data <strong>and</strong><br />
Simulation Results<br />
53