pharmacometrics: a shot in the arm for vaccine …
TRANSCRIPT
PHARMACOMETRICS: A SHOT IN THE ARM FOR VACCINE DISCOVERY AND DEVELOPMENT
~OR~VACCINES ARE NOT IMMUNE
TO THE CHARMS OF PHARMACOMETRICSPAGE 2019 Meeting, Stockholm, Sweden
Jeff SachsPharmacokinetics, Pharmacodynamics, and Drug Metabolism – Quantitative Pharmacology and Pharmacometrics, Merck & Co., Inc., Kenilworth, NJ, USA
June 12, 2019
Outline
• Motivation – Impact of vaccines• Potential impact for PMXMotivation
• Vaccines & ImmunologyBackground• Applications• Their impactPMX
2
Many Diseases Have Been Prevented
3
Halsey, NA, How Vaccines Cause Adverse Events, ADVAC Course Annecy France 2018
https://www.forbes.com/sites/matthewherper/2013/02/19/a-graphic-that-drives-home-how-vaccines-have-changed-our-world/ - 1a58e6dd3302
Age < 5Age ≥ 5
Polio
4
Historical Perspective: Smallpox
5
Bazin, H., Vaccination: a History
Don’t Count Your Children Until The Measles Have Come Through– African saying
6
20.3 Million*
*MMWR / November 11, 2016, 65 (44) 1228-1233
The Modern Toll of Measles
EU region2018
83,000 cases50,000 hospitalized
72 deaths1
Philippines1Q 2019
33,000 cases> 450 deaths3
7
1 European Region statistics. From “Measles in Europe: record number of both sick and immunized,” WHO Regional Office for Europe, Copenhagen, 7 February 20192 http://www.euro.who.int/en/media-centre/sections/press-releases/2019/over-100-000-people-sick-with-measles-in-14-months-with-measles-cases-at-an-alarming-level-in-the-european-region,-who-scales-up-response3 https://www.npr.org/2019/05/19/724747890/measles-outbreak-in-the-philippines
EU region1Jan18 – 8May19
100,000 cases>90 deaths2
Health and Economic Impact of Preventing Disease with Vaccination…
Just One Country: USAJust One Year’s Cohort: Children born in 2001
8
Adapted from: Zhou F et al. Arch Pediatr Adolesc Med. 2005;159:1136–1144.
All costs are given in US dollars (USD).Direct program costs included vaccines, administration, parent travel, and direct costs for the management of adverse events. Societal costs included direct program costs and parent time lost for vaccination and the management of adverse events.
>13 Million Cases Prevented > $53 Billion Saved
About 50 Vaccines Developed to Date
9
Adapted from: IOM (Institute of Medicine), Ranking vaccines: A prioritization framework: Phase I: Demonstration of concept and a software blueprint. Washington, DC, The National Academies Press, 2012, p. 19.
Number of Vaccines Developed:~30 as of 1990, ~50 as of 2010
1800 1900 2000
Cumu
lative
Num
ber
30
50
Smallpox10
…But Expensive and Takes Too Long
• Cost of a vaccine from discovery through Ph. 2a: $0.4 Billion (range $0.1-1B)*
• Time for a vaccine from discovery through Ph. 2a: 7 years (range 4-15 years)*
• Vaccines too often in development for ~20 years
10
Pharmacometrics
*Gouglas, D., TT Le, et al., Estimating the cost of vaccine development against epidemic infectious diseases: a cost minimisation study, Lancet Glob. Health, 2018;6:e1386–96
(Typically)• In Phase 3, 18,000 subjects
total, 3 years (range 1-5)
• Time: enrollment + low incidence rate
“Vaccine”
11
Vaccine (for today): Active Stimulator Of Immune Memory and Antibody Production for Prevention of an Infectious Disease.
12
Therapeutic
Chemo-Prophylaxis
PassiveImmunization
Seasonal Flu
Jeryl Lynn Hilleman with her sister, Kirsten, in 1966 getting the mumps vaccine developed by their father.
What is Special About Vaccines and Pharmacometrics?
Why were Vaccines not on our radar??• PK* Rare• Little DDI* (concomitant vaccination)• Traditional clinical pharmacology analyses not typical
• except safety & Tox not part of our traditional purview.
13*PK: pharmacokinetics, DDI: drug-drug interaction
The BASICS:VACCINES and IMMUNOLOGY
14
Active Immunization – How it works
• Measuring Immune response: “Titer” ~ Target engagemento Quantity and quality of antibodies
• More is better
15
Immunogenicity ≠ efficacy
Immune Memory
Recognize pathogen
Ramp up defenses quickly
Reduce symptomsor prevent disease
Overview of PMX and Vax
16
Vaccine Pharmacometrics
Modeling and Simulation in Vaccines
17
Computational vaccinology:
EpidemiologyHealth Econ
chem eng. & SYS BIO for bioprocess
PKPDof mAb
(“passive immu-
nization”)
QSP/PKPD in Chemo-
prophylaxis
immuno-genicity
= F(antigensequence)
Vaccine Pharmacometrics
(Today’s focus)
Rich History of Published Work (not a complete list!)
18
Hobson D, et al. (1972) The role of serum haemagglutination-inhibiting antibody in protection against challenge infection with influenza A2 and B viruses. J Hyg Camb. 70:767.
Fraser C, et al. (2007) Modeling the long-term antibody response of a human papillomavirus (HPV) virus-like particle (VLP) type 16 prophylactic vaccine. Vaccine. 25:4324–4333.
Coudeville L, et al. (2010) A new approach to estimate vaccine efficacy based on immunogenicity data applied to influenza vaccines administered by the intradermal or intramuscular routes. Hum Vaccin. 6(10):841–848.
Desai K, et al. (2012) Modelling the long-term persistence of neutralizing antibody in adults after one dose of live attenuated Japanese encephalitis chimeric virus vaccine. Vaccine. 30:2510–2515.
Andraud M, et al. (2012) Living on Three Time Scales: The Dynamics of Plasma Cell and Antibody Populations Illustrated for Hepatitis A Virus. PLOS Computational Biology. 8(3).
Le D, et al. (2015) Mathematical modeling provides kinetic details of the human immune response to vaccination. Front Cell Infect Microbiol. 4:177
Gómez‐Mantilla JD, et. al. ADME processes in vaccines and PK/PD approaches for vaccination optimization. In: ADME and tPK/PD. Wiley; 2016.
Rhodes SJ, et al. (2016) The TB vaccine H56 + IC31 dose-response curve is peaked not saturating: Data generation for new mathematical modelling methods to inform vaccine dose decisions. Vaccine. 34(50):6285-6291.
Rhodes, SJ, et al. (2018) Using vaccine Immunostimulation/ Immunodynamic modelling methods to inform vaccine dose decision-making. npj Vaccines. 3:36.
Rhodes SJ, et al. (2019) Dose finding for new vaccines: The role for immunostimulation/immunodynamic modelling. J Theor Bio. 465:51-55.
18
Barbarossa, et al. (2015). Mathematical models for vaccination, waning immunity and immune system boosting: a general framework. arXiv:1501.03451.
Atcheson E, et al. (2019) A probabilistic model of pre-erythrocytic malaria vaccine combination in mice. PLoS One. 14(1):e0209028.
Modelling dose responses following vaccination: lessons from the PK/PD field Rockville, Maryland, May 29, 2014Organizers:
BMGF: Steve Kern LSHTM: Richard White, Sophie Rhodes Imperial College: Gwen Knight AERAS: Tom Evans, Lewis Schrager
Modelling dose responses following vaccination: lessons from the PK/PD field Rockville, Maryland, May 29, 2014Organizers:
BMGF: Steve Kern LSHTM: Richard White, Sophie Rhodes Imperial College: Gwen Knight AERAS: Tom Evans, Lewis Schrager
www.lshtm.ac.uk/research/centres-projects-groups/isid#welcome
EXAMPLES
19
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
20
Enabling Translation
Understanding TE* Biomarkers Using Past Data
Leveraging TE* Biomarkers for Trial Design
Making Earlier Decisions with Quantified Risk
Supporting Regulatory Interaction
*TE: Target Engagement (stimulating protective immune response)
Seven Examples
Capability DemonstrationProgram Impact
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
1/2What N (# subjects)
will let us tell if vaccines A and B are different?
Phenomenological model Clinical trial simulation
Trial design, program strategy for sequence of trials
21
Wrong Question!
What can we learn about dose-level & formulation impact on immunogenicity?How can we use past data to inform the trial design?How can we integrate data across trials in the future?
22
How many arms (and which ones) were needed to address information desired in the first question?
Number of arms the team had planned:
Trial Design by Phenomenological Simulation
Formulation 2
Low Med HighLow N12Med N15High N18
Formulation 1
Low Med HighLow N1 N2 N3Med N4 N5 N6High N7 N8 N9
Dose
of
Antig
en se
t B
Dose of Antigen set ASimulate
each of the 18 possible arms
Estimate power to detect dose /
formulation effects
Powe
r (%
)
High dose increases response by factor on x-axis
0 2 4 6 8 10 12 14 16 18 20-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Formulation 1 Formulation 2
Log
(titer
)
Simulation Trial Arm #
Comp
arato
r
Choose trialarms & sizes
Formulation 2
Low Med HighLowMedHigh
Formulation 1
Low Med HighLow NMed N NHigh N
Dose
of
Antig
en se
t B Dose of Antigen set A
Covariance of Clin. Immune response &
form of R = R(Dose, Formulation)
Thanks: Kapil Mayawala, Jon Hartzel
23
Trial Design by Phenomenological Simulation
Formulation 2
Low Med HighLow N12Med N15High N18
Formulation 1
Low Med HighLow N1 N2 N3Med N4 N5 N6High N7 N8 N9
Dose
of
Antig
en se
t B
Dose of Antigen set ASimulate
each of the 18 possible arms
Estimate power to detect dose /
formulation effects
Powe
r (%
)
High dose increases response by factor on x-axis
0 2 4 6 8 10 12 14 16 18 20-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Formulation 1 Formulation 2
Log
(titer
)
Simulation Trial Arm #
Comp
arato
r
Choose trialarms & sizes
Formulation 2
Low Med HighLowMedHigh
Formulation 1
Low Med HighLow NMed N NHigh N
Dose
of
Antig
en se
t B Dose of Antigen set A
Covariance of Clin. Immune response &
form of R = R(Dose, Formulation)
Thanks: Kapil Mayawala, Jon Hartzel
24
Impact: • changed from 2-arm “yes-no” study to 5-arms (same
total # subjects), optimized to learn key properties • Helped plan:
(1) next studies, and(2) how to integrate data across studies
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
2 What disease assay(s) should we use and how often?
QSP and Bayesian probabilistic
Saved $, increased POSChoice of assay, frequency
25
• Measuring vaccine efficacy requires counting number of disease cases.
• What happens if the counting (assay) process is not perfect?
• What assays should we use and how often?
Placebo Vaccinated
Efficacy = Proportional Risk Reduction10% of placebo subjects get sick,3% vaccinated subjects get sick
Efficacy = (10% - 3%) / 10% = 0.7 = “70% efficacy”
26
Placebo Vaccinated
Efficacy = Proportional Risk Reduction10% of placebo subjects get sick,3% vaccinated subjects get sick
Efficacy = (10% - 3%) / 10% = 0.7 = “70% efficacy”
27
Typically need tens/hundred cases (Ph. 2/3, resp.) No efficacy information until Ph. 2b/3
10% of placebo subjects get sick,3% vaccinated subjects get sick
Efficacy = (10% - 3%)/10% = 0.7 = “70% efficacy”Efficacy = (13% - 6%)/13% = 0.54 = “54% efficacy”
Efficacy = Proportional Risk Reduction
28
Placebo Vaccinated
Daniel Rosenbloom, Nitin Mehrotra
Impact of False Positives
+3%+3%
Repeat testing Smart testing
Ph. 2/3 DesignMore assays:
Trade Time for powerDuration of
detectability matters
+ Pathogen Dynamics
13%6%
Radh
a Ra
ilkar,
Dan
iel R
osen
bloom
, Cas
ey D
avis
Shorter-duration infection
Longer-duration infection
10% of placebo subjects get sick,3% vaccinated subjects get sick
Efficacy = (10% - 3%)/10% = 0.7 = “70% efficacy”Efficacy = (13% - 6%)/13% = 0.54 = “54% efficacy”
Efficacy = Proportional Risk Reduction
29
Placebo Vaccinated
Daniel Rosenbloom, Nitin Mehrotra
Impact of False Positives
+3%+3%
Repeat testing Smart testing
Ph. 2/3 DesignMore assays:
Trade Time for powerDuration of
detectability matters
+ Pathogen Dynamics
13%6%
Radh
a Ra
ilkar,
Dan
iel R
osen
bloom
, Cas
ey D
avis
Shorter-duration infection
Longer-duration infection
Impact:• Saved $, • increased POS (per subject)• Choice of assay(s) and their frequency
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
(na) Can we model enough of the immune system to be predictive? QSP Capability development
30
How can we use preclinical data to help design vaccines and to predict the right dose-level or regimen?
Basic Model of Some Immune System Components
31
Chen et al., CPT PSP 2014
Dendritic cells
T cells
Antigen presentation
B cellsPlasma cells
PK
Mechanistic QSP
2 6 10Time (months)
0
2
4
100
200
Antib
ody l
evel
Ab tit
ers
Thanks: Jeff Perley, Josiah Ryman
300
Basic Model of Some Immune System Components
32
Chen et al., CPT PSP 2014
Dendritic cells
T cells
Antigen presentation
B cellsPlasma cells
PK
Mechanistic QSP
2 6 10Time (months)
0
2
4
100
200
Antib
ody l
evel
Ab tit
ers
Thanks: Jeff Perley, Josiah Ryman
300Impact: Capability Development
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
4How many doses of vaccine are
needed to confer lasting protection?
QSPSuggests single dose could provide
protective immune memoryMechanistic insight
33
Can we leverage mechanistic information to help inform
regimen?
Hepatitis B: Models for Antigen, Anti-viral Titer and Immune Memory
34
Antigen
Immune memory
Anti-viral titer
• 10,815 anti-viral titres in 1,923 patients• 2-4 vaccinations in 6-48 month period• No Immune memory (Mi) or antigen (Vi) measurements
Wilson JN, Nokes DJ, Medley GF, Shouval D. Mathematical model of the antibody response to hepatitis B vaccines: implications for reduced schedules. Vaccine. 25(18):3705-12. 2007
35
• The model demonstrates significant differences between different vaccines in both the time taken to generate immune memory and the amount of memory generated.
• The model provides theoretical support for the hypothesis that a single vaccine dose can generate protective immune memory.
Authors’ Conclusions
Simulation Results
30Time (days)
0
24
6
8
Memo
ry (M
, mlU
/ml) 10
12
90 150 210 270 33030Time (days)
0
6
10
15
Vacc
ine an
tigen
(V, ų
g)20
25
90 150 210 270 330
Wilson JN, Nokes DJ, Medley GF, Shouval D. Mathematical model of the antibody response to hepatitis B vaccines: implications for reduced schedules. Vaccine. 25(18):3705-12. 2007
36
• The model demonstrates significant differences between different vaccines in both the time taken to generate immune memory and the amount of memory generated.
• The model provides theoretical support for the hypothesis that a single vaccine dose can generate protective immune memory.
Authors’ Conclusions
Simulation Results
30Time (days)
0
24
6
8
Memo
ry (M
, mlU
/ml) 10
12
90 150 210 270 33030Time (days)
0
6
10
15
Vacc
ine an
tigen
(V, ų
g)20
25
90 150 210 270 330
Wilson JN, Nokes DJ, Medley GF, Shouval D. Mathematical model of the antibody response to hepatitis B vaccines: implications for reduced schedules. Vaccine. 25(18):3705-12. 2007
Impact:• Suggests single dose might be
enough (for immune memory for this pathogen and vaccine mechanism)
• Mechanistic insight
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
2Which regimens should be tested
in Ph. 2 Trial?(Regimen: dose‐level, # doses, timing)
QSP Ph. 2 Trial Design: add new dose level and different regimen
37
Can we leverage mechanistic information to help inform
regimen in Ph. 2 trial?
NHP Data
Trial Design by QSP
38
• Predicted response to different dose-levels, regimens, formulations
• Predictions qualified with Ph1 data
• Changed Ph. 2 design to incorporate a lower dose-level, additional regimen
Thanks: Jeff Perley, Guido Jajamovich, Jos Lommerse (Certara), April Barbour
Literature Data(mostly non-clinical)
Clinical Data Overlaid on Translated Prediction
2Time (months)
101
102
103
104
104
Neutr
alizin
g Anti
body
Titer
s
105
4 6 8 10 12
106
0
200Time (Days)
2
3
4
Antib
ody T
iter (
log)
4000
Dose 300
200Time (Days)
4000
Dose1000
Antib
ody T
iter
Immune Cells1
Immune Cells2
Immune Cells3
Ab titer
NHP Data
Trial Design by QSP
39
• Predicted response to different dose-levels, regimens, formulations
• Predictions qualified with Ph1 data
• Changed Ph. 2 design to incorporate a lower dose-level, additional regimen
Thanks: Jeff Perley, Guido Jajamovich, Jos Lommerse (Certara), April Barbour
Literature Data(mostly non-clinical)
Clinical Data Overlaid on Translated Prediction
2Time (months)
101
102
103
104
104
Neutr
alizin
g Anti
body
Titer
s
105
4 6 8 10 12
106
0
200Time (Days)
2
3
4
Antib
ody T
iter (
log)
4000
Dose 300
200Time (Days)
4000
Dose1000
Antib
ody T
iter
Impact:• Increased confidence in ability to
• model regimen-response• Translate from non-clinical species
• Changed planned Phase 2 • dose-levels • number of doses
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
2,3Do we have adequate evidence of efficacy if some pathogens have too
few cases?
PoDBA ( = Probability of Disease
Bayesian Analysis)
Novel Ph. 3 endpointGNG test criteria
40
Can we increase POS by mitigating risk of a season with
low incidence rate?
PoDBA Method
• Estimate relationship between probability of disease and antibody titer values based on titer values of subjects with and without disease
41
Log Titer0
Prob
abilit
y of
Dis
ease
And
Tite
r Den
sity
Prob
abilit
y of
Dis
ease
Antibody Titer Value
vaccineplacebo
• Use this relationship and titer values of control and vaccine groups to estimate vaccine efficacyand its confidence interval
Prob
abilit
y of
Dis
ease
Antibody Titer Value
Wait for enough cases
Needmanycases
Method: Estimating the “Probability of Disease” Curve
• Use titer values measured in infected and non-infected subjects• Assume that the relationship between titer values and probability of disease follows a sigmoidal curve• Estimate the parameters of the curve and their confidence intervals using standard statistical method (Maximum likelihood)
log2 (Titer value)
Cou
nt
alldiseased
sample ofnon-diseased
Prob
abili
ty o
f dis
ease
log2 (Titer value)ET50
Pmax
slope
Pmax/2
0
Maximum likelihood estimate
Observed Estimated
42
Qualified PoDBA method & Efficacy CI (demonstrated predictive power) with published data and simulation
PoDBA Novel Endpoint (example from a program plan, not agency guidance on different approaches to basis of licensure)
43
case-count alone
+PoDBA
YesYesTrial P001
No
No
No correlate
(With correlate)
TBD
TBD
Yes
Impact:• Increased capability to develop “Correlate
of Protection”• Developed potential Correlate of Protection• Increased Power of planned trials
increased POS for same # subjects• Method for early decision and
understanding of covariates
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
2Is our immunogenicity likely to
provide the necessary protection, and at what dose‐level?
NLME+ MBMA + PoDBA(Comparator modeling +
PoDBA)No‐go, GO, Dose‐selection
44
Can we leverage literature data and early immunogenicity data to drive early, objective, risk-
based decisions?
Incid
ence
Rat
e
Dose-level
Incid
ence
Rat
e
Titer
Modeling Overview for Supporting Both Go and No-Go Decisions
1.Titer Incidence Rate (“IR”)• Published clinical data• Incidence rate for different disease levels• Data cover various populations
Dose-level Titer Incidence RatePredicted change in incidence rate efficacy
Tite
r
Dose-level
Vaxpbo
Efficacy =% Drop
3.Combine 1 & 2 “integrated” modeling:
2.Dose-level Titer• Relate dose-level to serum neutralization titer response• FIH Data
45
Visualization Has to Tie Together Data for Different Disease Levels and Populations
Titer
Incide
nce
rate
in Hu
mans
(%/ye
ar)
Population
Di
seas
e se
verity
leve
l
46
Titer
Incide
nce
rate
in Hu
mans
(%/ye
ar)
Pbo VaxA B C
12
3
Predicted biomarker-incidence rate relationship for population B at disease severity level 2 with 95% prediction interval
Data point weight 246
PopulationABC
Disease severity level123
Visualization Has to Tie Together Data for Different Disease Levels and Populations
Titer
Incide
nce
rate
in Hu
mans
(%/ye
ar)
Population
Di
seas
e se
verity
leve
l
47
Titer
Incide
nce
rate
in Hu
mans
(%/ye
ar)
Pbo VaxA B C
12
3
Predicted biomarker-incidence rate relationship for population B at disease severity level 2 with 95% prediction interval
Data point weight 246
PopulationABC
Disease severity level123
Pr( efficacy > E ) > T
E = sufficient for intended public health impactT = tolerance of organization for risk
Visualization Has to Tie Together Data for Different Disease Levels and Populations
Titer
Incide
nce
rate
in Hu
mans
(%/ye
ar)
Population
Di
seas
e se
verity
leve
l
48
Titer
Incide
nce
rate
in Hu
mans
(%/ye
ar)
Pbo VaxA B C
12
3
Predicted biomarker-incidence rate relationship for population B at disease severity level 2 with 95% prediction interval
Data point weight 246
PopulationABC
Disease severity level123
Pr( efficacy > E ) > T
E = sufficient for intended public health impactT = tolerance of organization for risk
Impact:Objective, Quantitative…
No-Go: $40M trialGo: $10M trial
Dose-level, Biological insight, …
Key question, Method, Decision / ImpactPhase Key Question Method(s) Decision(s) / Impact
2What N (# subjects)
will let us tell if vaccines A and B are different?
Phenomenological model Clinical trial simulation
Trial design, program strategy for sequence of trials
2 What immunogenicity assay(s) should we use and how often?
QSP and Bayesian probabilistic
Saved $, increased POSChoice of assay, frequency
(na) Can we model enough of the immune system to be predictive? QSP Capability development
4 How many doses of vaccine are needed to confer lasting protection? QSP Suggests 1 less dose
Mechanistic insight
2Which regimens should be tested
in Ph. 2 Trial?(Regimen: dose‐level, # doses, timing)
QSP Ph. 2 Trial Design: add new dose level and different regimen
2,3 Do we have adequate evidence of efficacy if some pathogens have too few
cases?
PoDBA ( = Probability of Disease
Bayesian Analysis)
Novel Ph. 3 endpointGNG test criteria
2Is our immunogenicity likely to provide the necessary protection, and at what
dose‐level?
NLME+ MBMA + PoDBA(Comparator modeling +
PoDBA)No‐go, GO, Dose‐selection
49
Vaccine Pharmacometrics
Today• Simulation-based trial design to add/save
trial arms or subjects• QSP modeling of the immune system as a platform• Using dose, regimen, formulation to predict
immunogenicity/efficacy• Translation between preclinical and
clinical immunogenicity• Leveraging literature data• Establishing new trial endpoints• Understanding covariate (age, genetics, geography,…)
effects on immunogenicity & efficacy
Don’t forget also…• Predicting most effective vaccine platform by
mechanistic modeling• Understanding or predicting safety/toxicity• Predicting the best route of administration• Leveraging results of real-world trials• Prioritizing vaccine candidates• Prioritizing pathogen candidates
50
CONCLUSION
51
52Source: https://www.ag.ndsu.edu/news/columns/beeftalk/beeftalk-an-ounce-of-prevention-is-worth-a-pound-of-cure/.
53
NSAID Anti-viral
Anti-biotic
54
• Vaccines are a key component of public health• Vaccines are a key component of public health• Pharmacometrics useful for vaccine discovery & development
‐ QSP, PK/PD, Bayesian, comparator, translational,…• These (and other) methods have impacted decisions• Pharmacometrics (we) can impact human health by helping
inform vaccine discovery & development‐ Assumptions, study design & strategy, data interpretation
Acknowledgements• Subjects (& their Parents)
• MSD– Discovery & Development Teams, PPDM-Bioanalytics– Vaxmodsim team+:
Luzelena Caro, Carolyn Cho, Julie Dudasova, Pavel Fiser, Jon Hartzel, Sean Hayes, Justina Ivanauskaite, Regina Laube, Brian Maas, Nitin Mehrotra, Jeff Perley, Seth Robey, Radha Railkar, Daniel Rosenbloom, Josiah Ryman
– Paula Annunziato and Daria Hazuda– Dinesh de Alwis & Vikram Sinha, Nancy Agrawal– PPDM-QP2– Mike Pish & co. @ MCS– Skip Irvine, John Grabenstein
• Also…– Certara, Inc: Jos Lommerse, Nele Mueller-Plock, Michelle Green, Amy Cheung many others– Former MSD: Matt Wiener, Sandy Allerheiligen
55
Backup
PROPRIETARY ICONS HERE 56
Attributes of Vaccine Development
1. Stringent safety2. Trial size and duration (event frequency)3. Efficacy = proportional risk reduction4. Surrogate marker (“Correlate of Protection,” a.k.a. “CoP”) challenges5. Lack of translational models6. Need arm with placebo or active comparator7. Complex biologics, need to be transportable stable, usable
57
CoP Challenges: Knowledge, Time, Resources, Variability…
58
Knowledge• Which measurements?• Which species?• Predictive power?
Resources• Many samples• Often multiple species• Resource-intensive
assays
Time• Knowledge early enough• Timely availability of
clinical data
Variability• Assays often +/- 2-fold• Large BSV in response• Variability in CoP predictive
power? BSV: between-subject variability