COVID-19: Correlates of Protection

Home , Dan Barouch, Kizzmekia Corbett, Operation Warp Speed

Workshop hosted by the Clinical Development & Operations and Enabling Sciences SWAT Teams

Thursday, November 19, 2020 Agenda

Time (CET) Topic Speaker(s) 15:00 – 15:05 Welcome & Meeting Objectives Peter Dull, Ivana Knezevic Session 1: Evidence for Existence of an Immune Correlate for Covid-19 15:05 – 15:20 Correlates of Vaccine-Inducted Immunity – An Overview Stanley Plotkin 15:20 – 15:40 SARS-CoV-2 immunity overview and risk factors for re-infection Florian Krammer 15:40 – 15:55 PK/PD Considerations for SARS-CoV-2 Neutralizing Antibodies Andrew Charles Adams 15:55 – 16:10 Non-human primate (NHP) passive transfer and vaccine studies Dan Barouch 16:10 – 16:50 Panel Discussion Moderated by Karen Makar Session 2: Operational, Statistical and Regulatory Considerations for Covid-19 Immune Correlates 16:50 – 17:00 Opportunities for CoP Identification from Ongoing Phase III VE Studies Kristen Earle 17:00 – 17:15 Regulatory Perspective: Approach to Acceptance of CoP for Licensure Daniel Brasseur 17:15 – 17:25 COVID-19 Immunoassay Platform Overview Richard Koup Development of the COVID-19 Research Standards and 17:25 – 17:35 Valentina Bernasconi Global Immunoassay Network 17:35 – 17:50 Statistical Approaches for Assessment of Immune Correlates of Protection Peter Gilbert 17:50 – 18:25 Panel Discussion Moderated by Peter Dull 18:25 – 18:30 Wrap Up & Next Steps Paul Kristiansen, Jakob Cramer

Privileged and confidential 2 Welcome & Meeting Objectives Peter Dull Deputy Director, Integrated Clinical Vaccine Development (Bill & Melinda Gates Foundation)

Ivana Knezevic Group Lead, Norms and Standards for Biologicals (WHO)

3 Context for today’s workshop

• Recent positive results from large Covid-19 vaccine efficacy studies is great news but we need multiple licensed products to have the necessary near-term global impact

• Complexities for operational pathways for next vaccine registration highlights the urgent need to accelerate progress toward identification of an immune correlate of protection. A correlate could:

o Accelerate access to additional Covid-19 vaccines through alternative study designs

o Support evaluation for durability of protection for each vaccine

• There is currently limited and scattered evidence on COVID-19 immune correlates which will benefit from a consolidated review and dialogue

Privileged and confidential 4 Objectives for the first half of the workshop

• Review correlates nomenclature and highlight core principles in the approach to identification of immune correlates through key examples from past efforts

• Present the accumulated evidence for immune responses to coronaviruses including SARS-CoV-2

• Review efforts to identify an association of re-infection risk with antecedent immune profile including an overview of ongoing and planned studies

• Share available NHP mAb PrEP preclinical data dose response results as well as recent early treatment clinical data to inform contribution of targeted antibodies for protection

• Review NHP dose-titration protection data from convalescent sera and vaccine studies and discuss evidence for contribution of cell-mediated immunity to protection in pre-clinical models

Privileged and confidential 5 Correlates of Vaccine- Induced Immunity – An Overview Stanley Plotkin, MD Emeritus Professor of Pediatrics (University of Pennsylvania)

6 by Stanley A. Plotkin 1. Basic immunology 2. Enables correct choice of vaccine antigen 3. To permit consistency of potency 4. To determine susceptibility of an individual or a population 5. If efficacy trial not feasible or ethical, immunological data enable licensure of vaccine 6. Enables bridging from first-generation vaccine to second generation 8 Correlate of Protection (CoP): An immune response that is statistically interrelated with protection Absolute Correlate: A specific level of response highly correlated with protection: a threshold Relative Correlate: Level of response variably correlated with protection Co-Correlate: One of two or more factors that correlate with protection in alternative, additive, or synergistic ways.

9 An immune response that is responsible for protection

10 Formerly called: Surrogate: An immune response that substitutes for the true immunologic correlate of protection, which may be unknown or not easily measurable

11 1. Levels of passively administered or maternal antibody that protect 2. Analysis of immune responses in protected and unprotected subjects in efficacy trials 3. Observations made on vaccine failures, e.g. immunosuppressed individuals 4. Human challenge studies 5. Extrapolation from animal challenge studies, including immunodeficiency

12 Serum Antibody CD4+ T cells Neutralizing B cell help Non-neutralizing (ADCC, etc.) T cell help Functionality (opsonsphagocytosis) Th17) Avidity Cytokines Lysis Tregs Mucosal Antibody CD8+ T cells IgA locally produced Lysis IgG diffused from serum Avidity 13 Must Define Protection. Against what? Infection? (Local or Disseminated) Disease? (Mild or severe)

14 Protection against disease IgG serum antibodies

Protection against infection IgA- IgG mucosal antibodies

15 16 ▪ T cell deficient humans suffer serious and fatal measles. ▪ Monkeys vaccinated with measles HA alone (low CD4+ T cell response) are protected against rash, but remain chronically viremic. ▪ Monkeys depleted of CD8+ T cells have increased viremia.

Pan CH, PNAS, 2005 17 18 No. of pfu in challenge 10 100 1000

Group

Seronegative

Naturally seropositive

Vaccinated seropositive

Plotkin,S.A. et al. J Infect Dis 1989;159: 860-865. 19 20 21 Influenza

Coudeville, L Personal communication 22 23 Age ELISA % Bactericidal % Efficacy in (years) Pos.* Pos.# Canada 1 93 18 0% 2 94 35 3 92 56 41% 4 94 75 5 84 68 Adult 100 100 83%

* 2 mcg/ml #  1/8

Maslanka SE, et al. Infect Immun, 1998, 66:2453-59, DeWals P, et al. JAMA 2001 24 25 (Artificial Challenge in Children)

Serum Nasal HAI lgA Shedding - - 63% - + 19% + - 15% + + 3%

Belshe, JID, 2000 26 27 28 29 Varicella gE protein plus ASOI adjuvant Efficacy > 90%

▪ Vaccine boosts VZ antibody, which is used as an nCop, but mCop is VZ-specific CD4+ lymphocyte proliferation stimulation index ≥ 5.0

Hata et al. NEJM 2002 30 31 ▪ Mucus ▪ Innate immune responses ▪ Barrier to vascular entry ▪ Mucosal lgA Antibody ▪ Mucosal lgG Antibody ▪ Serum lgG Antibody ▪ T cell responses

32 ▪ Binding antibodies that prevent attachment (Ebola) ▪ Th17 that attract PMN’s and prevent carriage (pneumococcal, TB?) ▪ Antibody Dependent Cellular Cytotoxic antibody (HIV) ▪ Stimulation of CD4+ T helper cells that secrete cytokines (pertussis)

33 Jang, Y., Seong, B., Expert Opinion on Drug Discovery, 2020 34 Influenza Vaccines as Examples of Complexity of Correlates HAI titer is usually used as mCoP but Micro-neutralization and ELISA may be better HA Stalk antibodies contribute Antibodies to neuraminidase contribute ADCC antibodies do not contribute Mucosal lgA helps, at least for LAIV CD8+ T cell function important in the elderly

Christensen, S., et al, JVI April 2019; Ng, S., et al, Nat Med June 2019 35 POLYTHEISM IS PREFERABLE TO MONOTHEISM

36 Evidence for Existence of a Correlate: SARS- CoV-2 immunity overview and risk factors for re- infection Florian Krammer, PhD Professor of Vaccinology (Icahn School of Medicine at Mount Sinai)

37 Evidence for Existence of a Correlate: SARS-CoV-2 immunity overview and risk factors for re-infection

Florian Krammer Mount Sinai Professor in Vaccinology Icahn School of Medicine at Mount Sinai

COVAX CoP Workshop November 19th, 2020 SARS-CoV-2 antigens

PBD # 6VXX

Grifoni et al., Cell, 2020 Potential correlates of protection: Antibody responses

• Antibodies to spike/RBD are often neutralizing • Antibodies to NP, nonstructural proteins are non-neutralizing • Fc-FcR interactions have so far been shown to play a negligible role in protection

• IgM and IgA seem relatively short-lived • IgG response seems normal/long-lived

• Mucosal antibody is present after infection ELISA reactivity to spike protein and neutralization titers correlate

N = 120 Wajnberg et al., Science, 2020 How long-lived are these antibody responses?

Wajnberg et al., Science, 2020 Potential correlates of protection: Memory B cells

https://www.medrxiv.org/content/10.1101/2020.08.11.20171843v2 Potential correlates of protection: T-cells

CD4+

CD8+ What do we know about protection from human coronavirus (hCoV) infection? Historic challenge studies with 229E

Author Type of study Virus Result Correlate Callow et al., Challenge/re- 229E Challenge: 10 of 15 infected Antibody???? 1990 challenge Re-challenge: Partial protection from reinfection, protection from symptomatic infection Callow, 1985 Challenge 229E Asymptomatic and uninfected sIgA, neutralizing antibody individuals had much higher baseline neutralizing antibody titers Reed 1984 Challenge 229E 6/6 protected from re- Antibody? challenge with homologous virus (no virus, no symptoms) Barrow 1990 Challenge 229E Lower proportion of Antibody individuals with high antibody titer experience significant colds compared to group with lower titers Adapted from Huang et al., Nature Communications, 2020 ….but:

….durability may be limited and not cross protective Is there any evidence for correlates of protection against SARS-CoV-2 infection in humans? A glimpse of evidence for protection by neutralizing antibodies from a fishing vessel

• 122 individuals on the ship • 3 had neutralizing antibodies before going to sea • Outbreak with 82.5% attack rate occurred

Individuals with neutralizing antibodies were not infected PARIS/SPARTA PARIS (Protection Associated with Rapid Immunity to SARS-CoV-2)

Aim: To investigate if reinfection is possible and to determine the protective titer • PARIS follows individuals with and without antibodies to SARS-CoV-2 for at least one year • Target enrollment: • 400 individuals • 50% of the participants are positive, 50% negative • Serum/PBMCs collected every two weeks • Serology (research grad plus selected time points with CLIA ELISA) • Saliva collected every week • Additional screening with Biofire respiratory panel • Continued testing for SARS-CoV-2 by PCR • Initiated on April 27th 2020

In collaboration with Viviana Simon at Mount Sinai Initial antibody titers (ELISA)

L o n g i t u d i n a l a n t i b o d y l e v e l s

1 0 0 0 0

1 0 0 0

r u

Day 0 is c

e h

enrollment, not t

r e

symptom onset, d

n u study started in

a 1 0 0 e

the end of April, r A most people were likely already infected in March

1 0

0 7 1 4 2 1 2 8 3 5 4 2 4 9 5 6 6 3 7 0 7 7 8 4 9 1 9 8 1 0 5 1 1 2 1 1 9 1 2 6 1 3 3 1 4 0

d a y s p o s t b a s e l i n e Initial antibody titers (ELISA)

L o n g i t u d i n a l a n t i b o d y l e v e l s

1 0 0 0 0

1 0 0 0

Studye expanded to 9 sites across the

r u

Day 0 is c

e h

enrollment, not t

US with target enrollment of >10,000

r e

symptom onset, d

n u study started in

a 1 0 0 e

the end of April, r individuals A most people were likely already infected in March

1 0

0 7 1 4 2 1 2 8 3 5 4 2 4 9 5 6 6 3 7 0 7 7 8 4 9 1 9 8 1 0 5 1 1 2 1 1 9 1 2 6 1 3 3 1 4 0

d a y s p o s t b a s e l i n e PARIS (SEM CIVIC)/SPARTA (CIVR)

Commonalities between all sites: • Samples take every 2 months (most sites have shorter intervals) • Serum • Saliva • PBMCs (selected sites, but for several thousand subjects) • Common serology (Mount Sinai ELISA) • Nasal swap/nasopharyngeal sample take if somebody becomes symptomatic • SARS-CoV-2 PCR • Most sites also run a respiratory panel/Biofire • Primary analysis at sites • Secondary analysis: Sarah Cobey and Marc Lipsitch Similar initiatives

• Public Health England • DoD • CDC • Arnaud Marchant at National Public Health Institute, Belgium A major difference between infection and vaccination: mucosal immunity Another potential difference between infection and vaccination: Binding to neutralization titer ratios Binding Neutralization Natural Natural infection infection

Binding:Neutralization for natural infection → 7:1 Binding:Neutralization for natural vaccination →20:1-40:1 Walsh et al., NEJM, 2020 Ania Wajnberg Department of Microbiology/ (Mount Sinai Hospital) Acknowledgements Icahn School of Medicine at Mount Sinai Peter Palese Carlos Cordon-Cardo Adolfo Firpo (Mount Sinai Hospital) Daniel Stadlbauer Fatima Amanat Harm van Bakel (ISMMS)

Viviana Simon Mia Sordillo Maria Bermudez-Gonzalez David Rich Denise Jurczyszak Judy Aberg Matt Hernandez (Mount Sinai Hospital)

Adolfo García-Sastre Lisa Miorin Teresa Aydillo

[email protected] Tom Moran http://labs.icahn.mssm.edu/krammerlab/ Twitter: @florian_krammer Katherine Kedzierska (U Melbourne) Jussi Hepojoki (U Helsinki) Olli Vapalahti (U Helsinki) PK/PD Considerations for SARS-CoV-2 Neutralizing Antibodies Andrew C. Adams, PhD Vice President, New Therapeutic Modalities & COVID-19 Research (Eli Lilly and Company)

59 PK/PD Considerations for SARS-CoV-2 Neutralizing Antibodies

Andrew C. Adams, Ph.D., Vice President, New Therapeutic Modalities & COVID-19 Research Eli Lilly & Company Therapeutic Antibodies to Treat Viral Infection

Passive immunization as infectious disease treatment • Diphtheria anti-toxin • Convalescent plasma • Ebola, influenza Prior mAbs- prevention/treatment • Palivizumab- RSV 1 • Zmapp, MAb114, REGN-EB3- Ebola 2

Targeted mechanism: SARS-CoV-2 Block ACE2/Spike protein interaction by • 3 Clerkin et al: COVID-19 and Cardiovascular Disease. Circulation. 2020; 141:1648-1655). antibody specific to spike protein

1 https://www.accessdata.fda.gov/drugsatfda_docs/label/2014/103770s5185lbl.pdf 2 Mulangu et al: Randomized, Controlled Trial of Ebola Virus Disease Therapeutics. N Engl J Med 381, 2293-2303 (2019).

61 Discovery and Path to FHD

Jones et al, https://www.biorxiv.org/content/10.1101/2020.09.30.318972v1 • Single B cell screening of sample 20-day post symptom onset by multiplex bead and live cell assays

• Sequence and identification of antibodies, selection, cloning and expression

• Analysis (binding validation, functional validation, biophysical characterization, epitope binning.

• Discovery of Bamlanivimab

62 Non-clinical PK/PD to Inform Dosing: Rhesus Prophylaxis Model of SARS-CoV2 Infection Bamlanivimab or control was administered 24-hours prior to viral challenge, and virus was sampled from upper and lower respiratory tract through day-6 sacrifice Lower Respiratory Tract Upper Respiratory Tract

Jones et al, https://www.biorxiv.org/content/ 10.1101/2020.09.30.318972v1 = significant < 0.05 63 Risk Assessment for FIH Dosing

Target-Based Molecule Based Non-clinical Safety Models • IgG CDRs derived • Standard IgG1 • Standard Models Lack from B-cells of • Decades of industry Target naturally infected experience • Animal models of patient • No inherent safety disease were emerging, • Human IgGs have concerns for the but not well undergone natural platform characterized positive and negative selection

• Concluded there was a low degree of an inherent risk of target organ toxicity • Theoretical safety/efficacy concerns related to: • Antibody-dependent enhancement (ADE) of viral replication or disease • Development of resistance/escape mutations Lilly approach to selection of First In Human dose: Engage PK/PD and Pharmacometrics

Limited information in the literature given the nature of SARS-CoV-2 infection • There was a need to move as quickly as possible given the pandemic situation • This meant there was no pre-clinical data to inform human dose selection for candidate selection and FIH protocol design by March/April 2020

Therefore two methods were employed to increase the likelihood of finding an efficacious dose • Method 1: Viral dynamic PKPD-based modeling and simulation. The goal was to find the dose that results in the fastest time to viral clearance from infection or symptom onset • Method 2: Physiologically-based pharmacokinetic modeling. Takes into account the physicochemical properties of the drugs to predict human pharmacokinetics. The goal was to identify the dose that results in adequate drug concentrations (e.g., IC90 of viral neutralization) in patients over a 28 day period Viral Dynamic Modeling and Simulation - Step 1: Build a Viral Dynamic Model We developed a viral dynamic PKPD model-based simulator to rapidly screen multiple compounds to obtain general idea of the most effective doses based on multiple factors including: a) In vitro neutralization potency b) Drug distribution into lung c) Day of drug dosing d) Elimination rate of the virus in the absence of drug

This rapid screening approach is for an average individual. The next step was to incorporate inter-patient variability Viral Dynamic Modeling - Step 2: Include variability and select lowest dose

Simulated Doses Simulated Doses

Conservative assumptions in the modeling: - No effector function. Therefore less risk of underdosing if there is a benefit of effector function in the clinic. - Immune system effect on clearance of free virus assumed to be constant. In reality, the effect likely increases with time, therefore doses selected may be higher than needed. Physiologically-Based Pharmacokinetic Modeling (PBPK)

• Method uses physicochemical properties of the drug to determine Kuepfer, L., Niederalt, C., Wendl, T., Schlender, J.-F., Willmann, S., Lippert, J., Block, M., Eissing, T., & Teutonico, D. (2016). Applied Concepts in PBPK Modeling: How to concentration-time profile in various organs, including the lung Build a PBPK/PD Model. CPT: Pharmacometrics & Systems Pharmacology, 5(10), • Determined a dose (450 mg) would result in lung concentrations 516–531. above the in vitro IC90 in the majority of patients (>90%) over a day https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5080648/figure/psp412134-fig-0002/ 28 period Conclusions

• The two independent approaches gave a dose range of 450-700mg

• 700 mg was therefore expected to result in maximum efficacy

• Due to uncertainty in the translation of in vitro potency data to in vivo, a conservative approach was implemented to study higher doses to mitigate risk of underdosing

• The clinical trial results in Phase 2 now showed that 700 mg was at a plateau and all dose levels were equally effective Evidence for Existence of a Correlate: Non- human primate (NHP) passive transfer and vaccine studies Dan Barouch, MD, PhD Professor of Medicine (Harvard Medical School)

70 Correlate of Protection Against COVID-19 in Rhesus Macaques

Dan H. Barouch, M.D., Ph.D. Director, Center for Virology and Vaccine Research Beth Israel Deaconess Medical Center William Bosworth Castle Professor of Medicine Harvard Medical School Ragon Institute of MGH, MIT, and Harvard

COVAX Workshop on Correlates of Protection (Virtual) November 19, 2020 Two Critical Questions for COVID-19 Vaccine Development

• Is there natural protective immunity? Will individuals who recover from COVID-19 be protected against re-exposure?

• Is there vaccine-induced immunity? What are the immune correlates of protection? Chandrashekar et al. Science May 20, 2020; Yu et al. Science May 20, 2020 SARS-CoV-2 Infection Protects Against Re-Challenge in Rhesus Macaques

Primary Challenge Re-Challenge

9 79162 7162 8 78165 7165 7172 7172 7 77173 7173 6 76174 7174 7182 7182 5 75244 7244 4 74250 7250 7252 7252 3 M3ed ian Med ian 2 2 1 1 Log Viral Load in Lung in Load Viral Log 0 7 14 21 0 7 14 21 Days Following Challenge or Re-Challenge

Chandrashekar et al. Science May 20, 2020 Pseudovirus NAb Titers Correlate with Protection with Prototype DNA Vaccines

BAL Nasal Swab

a m 8 w

P<0.0001 8 P=0.0199

/ s

7 7

e R=-0.6877 R=-0.4162

e p

6 i 6

C o

5 5

4 A 4

N R

m 3 3

s g

2 2

o g

L 1 1 o

1.0 1.5 2.0 2.5 L 1.0 1.5 2.0 2.5 Log Pseudovirus NAb Titer Log Pseudovirus NAb Titer

Yu et al. Science May 20, 2020 Mercado et al. Nature July 30, 2020 Viral Loads (BAL)

Sham (N=20) S.PP (N=6) 7 B7S64 BS80 181267 6418 6 76145 7154 181302 7123 5 75130 7157 180662 7245 4 74261 171264 180689 191358 3 1391372 171219 181343 T423 2 T2436 T440 181363 T414 1 T1457 T463 Med ian Med ian Log sgRNA Copies / ml / Copies sgRNA Log 0 2 4 6 8 10 0 2 4 6 8 10

Days Following Challenge Viral Loads (BAL)

tPA.S (N=4) tPA.S.PP (N=4) S (N=4) 7 7 7 BM39 BO93 BS82 6 6 6

5 5BO49 5BO95 BV16 4 4 4 6855 7028 T796 3 3 3 2 26856 27056 T781

1 1Med ian 1Med ian Med ian 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 S.dCT (N=4) tPA.WT.S (N=4) S.dTM.PP (N=5) 7 7 7 BV95 BP11 BV58 6 6 6 BV96 5 5BV35 5BV79 6905 4 4 4 Log sgRNA Copies / ml / Copies sgRNA Log BV36 6770 6906 3 3 3 171170 2 26768 2T799 180215

1 1Med ian 1Med ian 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 Med ian

Days Following Challenge Viral Loads (Nasal Swab)

Sham (N=20) S.PP (N=6) 9 B9S64 BS80 181267 8 68418 7145 7154 181302 7 7123 7130 6 76157 180662 7245 5 75261 171264 180689 4 1491358 191372 171219 181343 3 T3423 T436 2 T2440 181363 T414 1 T1457 T463 Med ian

0 2 4 6 8 10 Med i0an 2 4 6 8 10 Log sgRNA Copies / Swab / Copies sgRNA Log Days Following Challenge Viral Loads (Nasal Swab)

tPA.S (N=4) tPA.S.PP (N=4) S (N=4) 9 9 9 8 8BM39 8BO93 BS82 7 7 7 6 6BO49 6BO95 BV16 5 5 5 6855 7028 T796 4 4 4 3 3 3 6856 7056 T781 2 2 2

1 1Med ian 1Med ian Med ian 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 S.dCT (N=4) tPA.WT.S (N=4) S.dTM.PP (N=6) 9 9 9 BV95 8 8BP11 8BV58 7 7 7 BV96 BV35 BV79 6 6 6 6905 5 5 5 BV36 6770 6906

Log sgRNA Copies / Swab / Copies sgRNA Log 4 4 4 3 3 171170 6768 3T799

2 2 2 180215

1 1Med ian 1Med ian 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 Med ian

Days Following Challenge

BAL Immune Correlates: Pseudovirus NAb l

Week 2 l Week 4

m m

/ /

R=-0.5267 s R=-0.6880 e

7 e 7

i i p

P=0.0007 p P<0.0001 o

6 o 6

C C

5 A 5

N N R

4 R 4

m m g

3 g 3

s s

2 g 2

o o

L L

1 1

k k a

1.0 1.5 2.0 2.5 a 1.0 1.5 2.0 2.5 3.0

e e P Log Pseudovirus NAb Titer P Log Pseudovirus NAb Titer

Nasal Swab Immune Correlates: Pseudovirus NAb b

b Week 2 Week 4

a a

w w

S S

/ /

R=-0.4919 R=-0.5839 s

s 9 9

e e i

i 8 P=0.0015 8 P<0.0001

p p o

o 7 7

C C

6 6

A A N

N 5 5 R

R 4 4

m m g

g 3 3

s s

2 2

g g o

o 1 1

L L

1.0 1.5 2.0 2.5 1.0 1.5 2.0 2.5 3.0

k k

a a e

e Log Pseudovirus NAb Titer Log Pseudovirus NAb Titer

P P Immune Correlates: NAb, ADNKA, ADCP

difference in average complete - partial z-score * protection (BAL and NS) 0 0.25 0.5 0.75 1 1.25 IgA S IgG1 S IgG2 S IgG3 S IgM S log NAb IgA RBD IgM RBD FcR2B S FcR3A S ADNP S ADCP S ADCD S IgG1 RBD IgG2 RBD IgG3 RBD FcR2A1 S FcR2A2 S FcR2A3 S FcR2A4 S FcR2B RBD FcR3A RBDADNP RBD ADCP RBD ADCD RBD FcR2A1 RBD FcR2A2 RBD FcR2A3 RBD FcR2A4 RBD ADNKA MIP1b S ADNKA CD107a S

P=0.0009 P=0.0092 P=0.0044 log NAb 2.8

50000 ADNKA MIP1b S S

2.4 b 40000 ADCP S

P A

I 25

M 30000 C log NAb & FcR2A3 S

2.0

o l

20 A

K 20000 log NAb & IgM RBD N

1.6 D 15 A 10000 log NAb & ADCD RBD

complete partial complete partial complete partial 0.00 0.25 0.50 0.75 1.00 AUROC

Galit Alter Dose-Reduction Ad26.COV2.S Vaccine Study

• Goal: To define threshold for protection and to assess for possible enhancement with subprotective responses

• Animals: Rhesus macaques (N=30)

• Week 0: Single-shot vaccination with Ad26.COV2.S • 1x1011 vp (N=5) • 5x1010 vp (N=5) • 1x1010 vp (N=5) • 2x109 vp (N=5) • Sham (N=10)

• Week 6: 105 TCID50 SARS-CoV-2 challenge (IN+IT) Pseudovirus NAb Titers

r 10000

A 1000

i v

o 100

e s

P 10 1011 5x1010 1010 2x109 Viral Loads (BAL)

1x1011 (N=5) 5x1010 (N=5) 7 7 6 6 5 5 Sham (N=10) 4 4 7 3 3 6 2 2 5 1 1 0 2 4 6 8 10 0 2 4 6 8 10 4 3 1x1010 (N=5) 2x109 (N=5) 2 7 7 1 6 6 0 2 4 6 8 10 5 5

4 4 Log sgRNA Copies / ml / Copies sgRNA Log 3 3 2 2 1 1 0 2 4 6 8 10 0 2 4 6 8 10

Days Following Challenge Viral Loads (Nasal Swab)

1x1011 (N=5) 5x1010 (N=5) 7 7 6 6 5 5 Sham (N=10) 4 4 7 3 3 6 2 2 5 1 1 0 2 4 6 8 10 0 2 4 6 8 10 4 3 1x1010 (N=5) 2x109 (N=5) 2 7 7 1 6 6 0 2 4 6 8 10 5 5

4 4 Log sgRNA Copies / ml / Copies sgRNA Log 3 3 2 2 1 1 0 2 4 6 8 10 0 2 4 6 8 10

Days Following Challenge Pseudovirus NAb Titers Correlate with Protection

BAL Nasal Swab

R=-0.7572 R=-0.8903

7 / 7

e s

i P<0.0001 P<0.0001

e p

6 i 6

c o

5 5

A A

N 4 4

R g

3 R 3

s g

2 2

l k

1 1

k e

1.0 1.5 2.0 2.5 3.0 a 1.0 1.5 2.0 2.5 3.0

p e

log pseudovirus NAb titer p log pseudovirus NAb titer Dose-Reduction Ad26.COV2.S Study

• Low-dose Ad26.COV2.S vaccine protected rhesus macaques in BAL (2x109 vp) and nasal swabs (1x1010 vp)

• Subprotective vaccine responses resulted in reduced protection but no evidence of enhanced disease

• NAb titers correlated with protective efficacy in three independent vaccine studies in rhesus macaques (N=117) IgG Adoptive Transfer Study

• Goal: To determine if purified IgG from convalescent macaques protects naïve animals and to define NAb threshold for protection

• Animals: Rhesus macaques (N=12)

• Day -3: Adoptive transfer of purified IgG • 250 mg/kg IV (N=3) • 25 mg/kg IV (N=3) • 2.5 mg/kg IV (N=3) • Sham IV (N=3)

• Day 0: 105 TCID50 SARS-CoV-2 challenge (IN+IT) Pseudovirus NAb Titers Following Adoptive Transfer

250 mg/kg (N=3) 25 mg/kg (N=3) 3 3

2 2

t i

T 1 1

-4 -2 0 2 4 6 8 10 12 14 -4 -2 0 2 4 6 8 10 12 14

b A

N 2.5 mg/kg (N=3) 0 mg/kg (N=3)

3 3

o L

2 2

1 1 -4 -2 0 2 4 6 8 10 12 14 -4 -2 0 2 4 6 8 10 12 14

Days Following Challenge Viral Loads (BAL)

250 mg/kg (N=3) 25 mg/kg (N=3) 7 7 6 6

5 5 l

4 4

3 3 s

e 2 2 i p 1 1

o 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

A 2.5 mg/kg (N=3) 0 mg/kg (N=3)

N 7 7 R

g 6 6 s

5 5 g

o 4 4 L 3 3 2 2 1 1 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

Days Following Challenge Viral Loads (Nasal Swab)

250 mg/kg (N=3) 25 mg/kg (N=3) 7 7 6 6

5 5 b

a 4 4 w

S 3 3

2 2

s e i 1 1

p 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

o C 2.5 mg/kg (N=3) 0 mg/kg (N=3)

A 7 7 N

R 6 6 g

s 5 5

g 4 4 o

L 3 3 2 2 1 1 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

Days Following Challenge Logistic Regression Analysis Define NAb Threshold Titer of Approximately 50 for Protection in This Model CD8 Depletion Study

• Goal: To determine the role of CD8 responses in protection against SARS-CoV-2 re-challenge in convalescent macaques

• Animals: SARS-CoV-2 convalescent rhesus macaques (N=10)

• Day -3: CD8 depletion with anti-CD8a mAb (MT807R1) • 50 mg/kg IV (N=5) • Sham IV (N=5)

• Day 0: 105 TCID50 SARS-CoV-2 re-challenge (IN+IT) Waning NAb Titers in Convalescent Macaques

4 P<0.0001

e t i P=0.0003

T 3

g 2

o L

1 e ) ) g 4 7 n k k e e e ll e e a w w h ( ( C e e - g g e n n R e e ll ll a a h h C C CD8 Depletion in Peripheral Blood

Sham (N=5) Anti-CD8 (N=5) 1000 1000

500 500 CD8+ / ul / CD8+

0 0 -4 -2 0 2 4 6 8 10 12 14 -4 -2 0 2 4 6 8 10 12 14

Days Following Re-Challenge Viral Loads Following Re-Challenge (BAL)

Sham (N=5) Anti-CD8 (N=5) 7 7 6 6 5 5 4 4 3 3 2 2 1 1

0 2 4 0 2 4 Log sgRNA Copies / ml / Copies sgRNA Log Days Following Re-Challenge Viral Loads Following Re-Challenge (Nasal Swab)

Sham (N=5) Anti-CD8 (N=5) 7 7 6 6 5 5 4 4 3 3 2 2 1 1 0 2 4 0 2 4

Log sgRNA Copies / Swab / Copies sgRNA Log Days Following Re-Challenge

Viral Loads Following Re-Challenge (Nasal Swab)

w S

7 P=0.0159

e 6

i p

o 5

A 4

N R

g 3

g 2

k 1 a

e Sham Anti-CD8 Naive P IgG Adoptive Transfer and CD8 Depletion Studies

• Purified IgG protects macaques against SARS-CoV-2 challenge in a dose-dependent fashion

• Threshold NAb titer for protection in this model are low ∼50

• CD8 depletion reduced protection against re-challenge in convalescent macaques with waning NAb titers

• These data suggest that NAbs alone protect, but cellular immune responses may also contribute when NAb titers are borderline or subprotective Acknowledgements Beth Israel Deaconess Beth Israel Deaconess Bioqual Ragon Institute Noe B. Mercado R. Keith Reeves Laurent Pessaint Carolin Loos Abishek Chandrashekar John D. Ventura Alex Van Ry Caroline Atyeo Jingyou Yu Kaylee Verrington Kelvin Blade Stephanie Fischinger Jinyan Liu Huahua Wan Amanda Strasbaugh John S. Burke Lauren Peter Mehtap Cabus Jared Feldman Katherine McMahan Janssen / J&J Renita Brown Blake M. Hauser Lisa H. Tostanoski Roland Zahn Anthony Cook Timothy M. Caradonna Xuan He Frank Wegmann Serge Zouantchangadou Aaron G. Schmidt Esther A. Bondzie Lucy Rutten Elyse Teow Galit Alter Gabriel Dagotto Rinke Bos Hanne Andersen Makda S. Gebre Danielle van Manen Mark G. Lewis MIT Emily Hoffman Jort Vellinga Doug Lauffenburger Catherine Jacob-Dolan Jerome Custers UNC Marinela Kirilova Johannes P. Langedijk David R. Martinez Funding Zhenfeng Li Ted Kwaks Ralph S. Baric BARDA/J&J Zijin Lin Mark J.G. Bakkers Ragon Institute Shant H. Mahrokhian David Zuijdgeest Children’s Hospital MassCPR Lori F. Maxfield Sietske K. Rosendahl Huber Yongfei Cai NIAID Felix Nampanya Paul Stoffels Bing Chen Ramya Nityanandam Mathai Mammen Joseph P. Nkolola Johan Van Hoof Shivani Patel Hanneke Schuitemaker Panel Discussion Moderated by Karen Makar, PhD Senior Program Officer, Discovery & Translational Sciences (Bill & Melinda Gates Foundation)

103 Discussion Panel Members and Potential Topics

Panel Members Potential Topics

• Meera Chand, MBBS, Consultant Medical 1. What lessons from past efforts to identify correlates are most Microbiologist (Public Health England) representative of the challenges we envision for Covid-19?

• Robert Seder, MD, Chief, Cellular Immunology 2. What are the limitations in using natural infection-induced protection Section, Vaccine Research Center (NIAID/NIH) on predicting protection from different vaccine platforms?

3. What are the key learnings from the pre-clinical models about the Prior speakers also available for questions & discussion: relative contribution of non-antibody mediated protection?

• Stanley Plotkin, MD, Emeritus Professor of 4. How well do the NHP models re-capitulate human disease and how Pediatrics (University of Pennsylvania) do we incorporate the differences in interpreting protection afforded • Florian Krammer, PhD, Professor of Vaccinology through different interventions (e.g., mAb's, convalescent sera, (Icahn School of Medicine at Mount Sinai) vaccine-induced immunity)?

• Andrew Charles Adams, PhD, Vice President, New 5. In the absence of a formally recognized CoP, what orthogonal Therapeutic Modalities & COVID-19 Research (Eli evidence would strengthen the argument that a certain level of Lilly and Company) neutralizing antibodies could be sufficient to protect from COVID-19? • Dan Barouch, MD, PhD, Professor of Medicine What evidence would weaken this? (Harvard Medical School) Privileged and confidential 104 SIREN: The impact of detectable anti SARS-COV2 antibody on the incidence of COVID-19 in healthcare workers Cohort enrolment update Design: Prospective longitudinal cohort of healthcare workers across the UK National Health 45000 106 NHS Trusts Service 40000 38,770 Participants* 35000 Primary Objective 30000 25000 To determine whether the presence of antibody to 20000 SARS-COV2 (anti-SARS-COV2) is associated with a 15000 10000 reduction in the subsequent risk of re-infection over the 5000 next year, by measuring reinfection in seropositive and Number ofParticipants 0 seronegative cohorts Other analyses

Incidence and prevalence of symptomatic and Week begining Participants asymptomatic SARS-CoV-2 infection in healthcare *26,937 directly into SIREN, 11,833 into SIREN associated studies workers by region Approximately 30% of the cohort with available baseline serology Characterisation of reinfection and persistent infection is seropositive. cases, incorporating Expanding the cohort ▪ Clinical characterisation ▪ Humoral and cellular immunity • October/November commence recruitment from Devolved Administrations (aiming to recruit ~15,000 participants); ▪ Viral genomics • Facilitate acute trust recruitment: 1. reducing antibody testing Cohort serology: Description of longitudinal serological frequency, 2. Guidance on pooling specimens, 3. Continued profiles, relationship between commercial assays and support from NHS staff testing programme neutralising activity • Start recruitment from Primary Care Detailed humoral and cellular immunity substudies Objectives for the second half of the workshop

• Review estimated timelines from OWS and other global Phase III efficacy studies and highlight critical study design elements and limitation in supporting immune correlates analysis

• Provide an overview of regulatory expectations for adequacy of evidence for an immune correlates to support product registration and discuss tools to enable progress when evidence is incomplete

• Discuss clinical immunoassays to support product registration of Operation Warp Speed candidates and review performance characteristics and interrelationship between assays

• Share information about the CEPI Centralized Laboratory Network and WHO standardization efforts, including the establishment of the ECBS-endorsed International Standard for anti-SARS-CoV-2 antibody

• Present an overview of the Operation Warp Speed statistical analysis plan and approach considering recent results of high short-term efficacy from interim analyses

Privileged and confidential 106 Latest results from Pfizer/BioNTech, Moderna, and Gamaleya

Platform • mRNA • mRNA • Ad26 >> Ad5 prime-boost Date of press release • November 18, 2020 • November 16, 2020 • November 11, 2020 Preliminary point • 92% (early data, not reaching statistical estimate of vaccine • 95% (p<0.0001) • 94.5% (p<0.0001) significance) efficacy • As of November 13, 2020, enrolled • On October 22, 2020, completed • Target enrollment of 40,000 participants 43,661 participants to date, 41,135 of enrollment of 30,000 participants who • As of November 11, 2020, >20,000 have Phase 3 study whom have received a second dose have received at least 1 dose enrollment been vaccinated with the first dose of the of the vaccine candidate vaccine and >16,000 with both the first and second doses of the vaccine • 170 cases (8 in vaccine group) • 95 cases (5 in vaccine group) • 20 cases Total number of cases • 10 severe cases (9 in placebo, 1 in • 11 severe cases, all in placebo group • No information provided on case severity vaccine group) • Safety data milestone required by US • Intends to submit data for an EUA with • Observation of study participants will FDA for Emergency Use Authorization US FDA in the coming weeks continue for six months after which the (EUA) has been achieved • Expects the EUA to be based on the final report will be presented Plans for licensure • Plan to submit within days to the US final analysis of 151 cases and a • The research data will be provided by FDA for EUA and share data with median follow-up of more than 2 months RDIF to national regulators of countries other regulatory agencies around the interested in purchasing the vaccine to globe streamline the registration process

Privileged and confidential 107 Opportunities for CoP Identification from Ongoing Phase III VE Studies Kristen Earle, PhD Program Officer, Vaccine Development & Surveillance (Bill & Melinda Gates Foundation)

108 Key Landscape and timing of early phase III VE trials that Interim analysis Primary analysis may contribute data to correlates analyses COR Potential correlates analysis

2020 2021 Developer Ph III Sites1 Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul

CanSino SAU, PAK, RUS Enrollment COR

Gamaleya RUS, BLR, UAE, VEN, IND Enrollment COR

UAE, PER, MAR, ARG, Sinopharm Enrollment COR BHR, JOR, EGY

Sinovac BRA, IDN, TUR Enrollment COR

Pfizer USA, ARG, BRA, GER, RSA Enrollment COR

Moderna USA Enrollment COR

BRA, UK, IND, RUS Enrollment COR Oxford / AZ USA Enrollment COR USA, BRA, ARG, CHL, COL, Janssen Enrollment COR MEX, PER, PHL, RSA, UKR RSA, UK, MEX Enrollment Novavax USA2 Enrollment Assumptions: How might we expedite? COVAX Clinical SWAT exploring options: • 6-month attack rate: • Primary analysis: 150 cases • US, UK: 2% • Recruitment / vaccination: 3 mo. • “Real time” analysis: Cases analyzed as they accrue • Others: 5% • Follow up for VE endpoint: 2 mo. • Minimize time between primary and correlates analyses • VE: 50% • Data mgt & analysis before IA and PA: 1 mo. • Pool data within platforms • Interim analysis: 75 cases • Preparation of correlates report: 2 mo.

1. Where developers are conducting multiple Phase III studies, timeline represents site with predicted earliest readout (bolded), based on public sources (primarily clinicaltrials.gov) and modeled assumptions. 2. Actual start date and study design TBC. Breadth of global response presents both challenges and opportunities to identification of a correlate

Some challenges unique to such a large-scale …but similarities in protocols may facilitate effort… analysis across studies

• Breadth of platforms: Humoral / cellular immunity • Timepoints: Sera collected in all subjects at contributions vary; unclear if universal correlate will apply baseline, 2-4 weeks following each dose, 6 mo. ILLUSTRATIVE • Inclusion / exclusion criteria: 18+, no prior history of laboratory-confirmed SARS-CoV-2, no fever Relative contribution to efficacy • Common assays: Binding titers (S, RBD, N ELISAs), wildtype and pseudovirus neutralizing assays • International standard: Research reagent available • Variability in antigens: Consensus around RBD / Spike, now, to be endorsed by WHO in December but stabilization / sequence may affect quality of Abs

Correlate must be derived from multiple VE trials, assessing multiple vaccine platforms Regulatory Perspective: Approach to Acceptance of CoP for Licensure Daniel Brasseur, MD, PhD Former EMA Expert

111 Regulatory Perspective BBBBApproach to Acceptance of CoP for Licensure Daniel Brasseur past CHMP and VWP member at EMA

Covid Vaccine Candidates: • Different Platforms/Concepts • Several close to ending Ph3 trials • Heterogeneous Target Populations

Sensitivity: CEPI Internal What information to be available in the near future?

Several candidates/platforms having shown:

• Clinical Efficacy using similar Primary Endpoints However, this endpoint might be no longer reachable at term (ethics, epidemics)

• Humoral Immune Response targeting ‘viral’ Spike antigen: Full length Spike protein, RBD…? However using different assays and ways to express data SNT IgG, GMTiters, GM fold rise, % > 4fold increase in IgG from baseline, cutoff threshold, reverse cumulative expression

• Additional antigens N, Matrix… Other assays?

Sensitivity: CEPI Internal What information might be missing in the near future ?

Several candidates/platforms not having ‘sufficiently’ addressed

• Specific populations (elderly, pregnancy, pediatrics…)

• Specific clinical conditions (co-morbidities, immuno-compromised…..)

• Specific regions (ethnic groups, pre-exposed populations)

• Targeted immune parameters (CD4/8, Cytokines, Interferon, Interleukines…)

• Duration of protection

Sensitivity: CEPI Internal FDA/NIH/CEPI Biomarkers Workshop September, 2019

“ Use of biomarkers for regulatory decision-making in Covid19 vaccine development and licensure application review... will be unavoidable ”

Sensitivity: CEPI Internal How to identify/select/confirm the Covid 19 Immune marker candidate for surrogacy ? Research • Detected convalescent serum during the recovery phase • Effective in passive transfer in animal model • Linked to protection in challenge studies (animal, humans?)

Development • ‘Responding’ in vaccine trials • Positively linked to protection in efficacy/effectiveness studies • Negatively linked to re-infections/breakthrough cases

Sensitivity: CEPI Internal MenC vaccine example shifting to…. AWY, & B

• SBA as a correlate of protection «defined» from natural history studies (Gotschlich et al., 1969)

• Accepted as CoP for Polysaccharide vaccines

• MenC-conjugate did not demonstrate clinical protection before licensing (Vipond et al., 2012)

• Plausible/convincing marker of protection : hSBA

• Significant/convincing immune response achieved in trials

• Confirmation in effectiveness (post-licensing studies)/breakthrough

• Extension bridging lab determinations (rSBA vs hSBA)

• Extension to similar platforms (change of carrier protein: Tet vs. CRM197)

• Extension to similar conjugates AWY

• Extension to OMV/protein-based vaccines: MenB Sensitivity: CEPI Internal Immune Marker reflecting directly the MoA of Protection • Qualify: functional IgG antibodies, (bactericidal with Complement) • Quantify: SBA titers (serum dilution linked to 50% killing at 60 min) allowing to make a long and beautiful trip from

Polyosidic Conjugate Protein Concept

Adult Pediatric Pediatric Target Population Ados C C …+ AWY + B Target Strains

CRM197 Formulations +TT+D hSBA rSBA or hSBA hSBA Lab determinations

Sensitivity: CEPI Internal Hepatitis B experience

• Vaccines protecting via IgG humoral response hepatitis B immune globulin (HBIG) protective when administered to neonates born to HBeAg-positive mothers)

• HepislavB response (Anti-HBs antibody levels ≥10 mIU/mL) non inferior to Engerix B in general population

• Adding an adjuvant to show immune superiority without jeopardizing the safety profile Fendrix non-inferior to EngerixB in ‘refractory’ renal patients

Sensitivity: CEPI Internal Pneumococcal conjugate experience

Protective threshold determined by relating observed antibody concentrations in populations studied in controlled trials to point estimates of efficacy

Pooling across multiple efficacy trials, valid because protective concentrations were similar across trials, obtained a more widely applicable level of 0.35 µg/ml

Source: Siber et al. (2007) Vaccine Sensitivity: CEPI Internal Regulatory Bridging Approach on ‘validated’ Surrogates of Efficacy

• Flu vaccines from the egg-based to cellular production, from HI to NA antigen, from HI, VN, SRH assay…

• Pertussis vaccines from the whole cell to the a-cellular vaccines

• Polio vaccines from the live Oral to the killed IM polio vaccines from tri-valent to mono-valent vaccines from ‘virulent’ to ‘attenuated’ production strains

• Pneumo Conjugates from one to several carriers ( CRM197, T, proteinD…) from 4 (clinically- KaiserP) to …another 9 (immunogically) studied serotypes

• HPV vaccines from mucosal (sIgA) to parenteral protection from 2 to several oncogenic serotypes

• …….

Sensitivity: CEPI Internal Regulatory Approach: a bunch of convergent evidence

A (non validated) CoP, hence a plausible surrogate immune marker • is never considered as a stand-alone decisive element but in context • together with a several different (limited clinical + supportive biological) arguments • since also potentially informing on

Protection • Cross reactivity? Cross protection? (strains as moving targets) • Immune memory (CMI playing the main role?)

Safety • Th1/Th2 polarization and enhanced disease • Immune response sustainability and (early) break through cases (need for a booster)

Sensitivity: CEPI Internal Bridging situations between Covid-19 platforms…

Considering bridging attempts to demonstrate ‘equivalence’ of the immune response elicited by 2 vaccines that were performed whether during: • Head-to-head trials • Re-runs of samples • Using same validated assay – For COVID-19 there is need for standardization • ….

In compliance with an appropriate and agreed methodology

What to request if facing the following scenarios: ????

Sensitivity: CEPI Internal Bridging situations between Covid-19 platforms…

Change within a given platform • Just another known adjuvant non-inferiority? • Just an additional antigen (S + …N, S1, S2, RDB, M) superiority (safety) non inferiority?

Comparing (conceptually) similar platforms • Another killed/sub-unit vaccine with identical Spike antigen Non inferiority? What margin? What confidence interval? • Another mRNA vaccine targetting the same S1antigen or RDB site NI + some CMI • Another vectored vaccine carrying identical antigen NI + Vector linked immune response?

Sensitivity: CEPI Internal Bridging situations between Covid-19 platforms… Adding Complexity and realistic insights

Comparing different concepts/platforms based on the anti-Spike antibody response, how to evaluate the protective contribution of CMI response?

• Licensed mRNA and candidate recombinant viral vectored vaccine Spike IgG Superiority and or comparable CMI responses? • Licensed mRNA and Candidate Sub-unit (adjuvanted) NI whilst anticipating stronger CMI responses? • Licensed Vectored A and Candidate vectored B (replicating vs non-replicationg) NI + ? Vector linked CMI impact…?

Sensitivity: CEPI Internal Conclusions

• Biomarkers are a critical part of vaccine discovery, development, licensure, and implementation. • Vaccine-associated biomarkers can include measures of immune response correlated with protection from disease AND nonimmune biomarkers related to safety and effectiveness. • From a regulatory perspective, biomarkers can be used for many different objectives. • Context of use (e.g., “traditional” approval, AA, etc.) is the critically important feature that informs and frames each individual case in which a biomarker is proposed to accomplish a regulatory objective.

Sensitivity: CEPI Internal COVID-19 Immunoassay Platform Overview Richard Koup, MD Deputy Director, Vaccine Research Center (NIH/NIAID)

127 Operation Warp Speed overview

Operation Warp Speed (OWS) is multi-agency USG effort conducted in collaboration with the private sector to protect the American people from future COVID-19 devastation by providing funding, coordination, and regulatory support to enable the rapid development and distribution of novel diagnostics, therapeutics, and vaccines

Diagnostics (Dx) Therapeutics (Tx) Vaccines (Vx) Stimulate innovation to Support pre-clinical and Accelerate clinical enable rapid, multi-modal clinical development, assessment, review of testing of COVID-19 infection manufacturing, and therapies as an adjunct to and immunity distribution/access future availability of vaccines Scope and mandate of the immune assay WG

OWS Vx development: John Mascola, MD Advisory Group DoD Lead: Matt Hepburn, MD

Preclinical assessment Immune assays Clinical trials Cristina Cassetti (NIH) Richard Koup (NIH) Merlin Robb (DoD) Clay Hughes (BARDA) Ruben Donis (BARDA) Mary Marovich (NIH)

Clinical development leading to To prioritize and down-select To establish immune assays for vaccine licensure candidate vaccines for clinical preclinical and clinical use • Conduct Pivotal phase 3 development • Binding assays (spike and other efficacy studies • Small animal immunogenicity domains) • Predictive modeling and protection • Pseudovirus and live virus neuts • Special population safety • NHP immunogenicity and • CD4 and CD8 T cells including children and women protection • Qualify, validate, tech transfer of child bearing age • VAERD and ADE in animal • Immune correlates analyses • Phase 1 studies of Co- models Focus of this group Administration / Mix & match • Human challenge contingency Overall charge • Define and deploy a core set of preclinical and clinical assays needed for key immunogenicity and efficacy endpoint evaluation

• Primary purpose will be for Immune Correlates of Protection OWS Immune assay working Elements • Define key assay needs (harmonized across pre-clinical and clinical) group charge – Binding antibody—to vaccine and non-vaccine viral antigens – Neutralization assays suitable for BSL-2 and BSL-3 environments – CD4+ and CD8+ T-cell assays

• Validate, qualify and tech transfer assays from a core set of laboratories to achieve redundancy and ensure capacity to support nonclinical & clinical testing

• Identify and source critical reagents

130 Development status of immune assays for Ph3

Assay lab Assay type Recommended assay Strategic rationale Validated by1 Capacity/week

MSD ECL: Spike (S2P), RBD Multiplex 384 well – very high and N - Multiplex Assay Binding throughput, greater dynamic range and (meso scale discovery - 2,400 uses less protein and sample than ✓ Antibody electrochemiluminescence standard ELISA assay)

Pseudovirus: Spike pseudovirus ~600 - Lentiviral vector, ffLuc reporter ✓ Redundancy of assays for risk Neutrali- Wild-type virus: WA isolate, mitigation December 500 zing passage 3, Vero-E6 cells Antibody Variable throughput, but all should be Reporter virus: Full-length sensitive and validatable molecular clone, Vero-E6 cells, January 500 nanoLuc reporter

High throughput and extensive Intracellular Cytokine Staining experience performed under GCLP in 625 T-cell (ICS) Assay clinical studies. High sensitivity and easy ✓ transfer of assay to other labs

1. Date is for the "initial" validation using convalescent samples; final validation is dependent on access to product (vaccine study-derived) samples and FDA review. Multiplexed OWS binding assay

Multiplex plate layout: • Simultaneous measurement of SARS CoV-2 Spike (S-2P), RBD, N • High throughput (384 well) – 20 samples/plate • Sensitive electrochemiluminescence detection • High sensitivity and precision • Large dynamic range • Conserves sample volume • No coating step needed

Assay plate Layout: • Standard Curve w/8 points in triplicates • SARS CoV-2 Convalescent sera control • Undiluted H/M/L controls (MSD) • Negative Controls in triplicates • 20 samples w/ 8 dilutions in duplicates • AU/ml and endpoint dilution results • IU/ml when NISBC standard is available (Dec, 2020)

Semi-automated: • Current throughput: 480 samples/day (Biomek i5) • Expansion to 6k/week (Biomek i7) Adrian McDermott, Vaccine Immunology Program – VRC-NIAID-NIH SARS-CoV-2 Spike Pseudotyped Virus Neutralization Assay in 293T/ACE2Vaccine core Cells team designed to support fast decision-making Lights “ON” Dx/Tx workstreams likely to share select resources with Vx, including single POC for BioPharma partners

MolecularVx Lead (Matt Hepburn) cloning & Other OWS workstreams Vice (Gary Disbrow)

Diagnostics Lead Administrative Chief of Staff & Program 293T/ACE2 Cell Support Mgmt

pLuc pTMPRSS2 Therapeutics pHIVEnv pSpike Vaccine development Supply/Manuf. Lead & Delivery & Admin. Lead Infection DNA + DNA Lead Portfolio Manager team & DoD lead DoD lead & DoD lead

Transfection

FNIH Consultant Private sector POCs

293T cell Other IO/support functions Pseudovirus Internal & external communications, Finance, Contracting, Legal (all TBD) • Adapted from Barney Graham, Kizzmekia Corbett, Nicole Doria-Rose, John Mascola- VRC, NIH David Montefiori, Duke University • 293T/ACE2 cells from Mike Farzan and Huihui Mou Live NanoLuc reporter virus assay

Ralph Baric, University of North Carolina Overview of Battelle’s SARS-CoV-2 Live Virus Microneutralization Assay

INFORMATION NOT RELEASED TO THE PUBLIC UNLESS AUTHORIZED BY LAW: This information to include the microneutralization assay to evaluate COVID-19 vaccines samples “in situ ELISA” and accompanying instruction s under provisional Patent Application No. 60/041,551 and Material Transfer Agreement executed between Battelle and NIAID has not been publicly disclosed and is privileged and confidential. It is for internal government use only and must not be disseminated, distributed, or copied to persons not authorized to receive the information.

135 Neutralizing antibody assays correlate with each other

Pseudovirus – 2 different labs Pseudovirus vs. Live virus

1 0 5 1 0 5 r2= 0.8728 r2= 0.7484

4 p <0.0001 4 p <0.0001 0

1 0 0 1 0

3 3

e 1 0

u D 1 0 2 D 1 0 2

1 0 1 1 0 1 1 0 1 1 0 2 1 0 3 1 0 4 1 0 5 1 0 1 1 0 2 1 0 3 1 0 4 1 0 5 H a rv a rd ID 5 0 U N C ID 5 0

David Montefiori, Duke University Redundancy of Laboratories for Capacity and Risk Mitigation

• Multiplex binding assay Fred Hutch • VRC • Battelle • CRO TBD • Pseudovirus neutralizing Ab assay • Duke • Battelle • Monogram Battelle Monogram VRC • Live virus neutralizing Ab assay • In situ ELISA • Battelle Duke • Microbiologics Microbiologics • CRO TBD • NanoLuc reporter virus UNC • UNC • Microbiologics • CRO TBD • ICS T cell assay • Fred Hutch Assay standardization across platforms (OWS)

• NIBSC International Standard o Live virus = 18 labs o Pseudovirus = 30 labs • Incorporate into MSD assay to allow IU/ml read-out • Lentivirus = 12 labs • VSV = 8 labs • In assays standards • Other = 10 labs • HML within MSD assay o Total = 48 labs • Large plasma -> serum panels Overall Results: ID80 of Positives • Vitalant, Battelle, SAB Biotherapeutics • Janet Lathey (NIAID) • EQA program • Pat D’Souza (DAIDS), Tom Denny (Duke) • SNACS program • David Montefiori, Duke

SNACS blinded labs Current prioritized vaccines

✓ ? ✓ ✓ ✓ ✓

mRNA Non-replicating Nucleic acid Viral vector Protein subunit

Platform type

Source: Vaccine Landscape Database, Operation Warp Speed Summary

• OWS Immune Assays Working Group has selected, qualified and (validated) multiple assays for use in comparative immunogenicity and immune correlates analyses in support of OWS phase 3 trials. • We are building capacity and redundancy through awarding contracts to CRO laboratories that will tech transfer and validate the assays from the developer laboratories. • We are working with multiple contractors and partners to identify and secure large volumes of sera for use as internal controls and reference standards for bridging, validation, trending, and external quality assessment. Thank You

Questions? Development of the COVID-19 Research Standards and Global Immunoassay Network Valentina Bernasconi, PhD Scientist, Preclinical and Immunology (CEPI)

142 The Coalition for Epidemic Preparedness Innovations Development of Covid-19 standard reagents and of a global immunoassay network Overview of CEPI Centralized Laboratory Network and WHO standardization effort, including upcoming international standard

Valentina Bernasconi

19th November 2020 COVID-19 vaccine development landscape

• More than 300 vaccine developers worldwide • Comparing immune responses against different vaccine candidates is challenging • Different stages of development • Biological variation • Technical differences (how and where specimens are collected, transported, stored, and analyzed) • Different technology platforms (e.g. recombinant viral vectors, inactivated vaccines, recombinant proteins and nucleic acids) • Lack of standardization among different assays and different testing laboratories

144 How to improve assay standardization?

1. Reference reagents at NIBSC, a WHO Collaborative Center • Research reagent and panels (Interim solution) • WHO International Antibody Standard established by the ECBS

2. CEPI Centralized Laboratory Network • Selection of laboratories with high quality standards worldwide • Selection of a core set of preclinical and clinical assays needed for key immunogenicity and efficacy endpoint evaluation • Harmonization of protocols and key reagents across the laboratories

145 Reference reagents at NIBSC Reference reagents and reference panel

• Create a communal language between laboratories worldwide and improve comparability of results

th 30th April 31st May 30th September 7th November 10 December

WHO public consultation

Convalescent serum samples Formulation of several Interlaboratory collaborative Results analysis Submission Establishment collection (from UK, candidates study and final report to WHO WHO International Singapore, Norway, US) ECBS Standard

Research Reagent and Reference panels

WHO International Antibody Standard

• To acquire the reference material: COVID-19-related research reagents available from the NIBSC

https://www.who.int/groups/expert-committee-on-biological-standardization 147 CEPI Centralized Laboratory Network CEPI Centralized Laboratory Network

The Netherlands

UK Canada Italy Porton Down, UK

USA India

Bangladesh

All laboratories in the Network can perform analysis of: - Binding antibodies - Neutralizing antibodies - T cells Objectives of the Network

The CEPI Centralized Laboratory Network is open to all COVAX funded and non-funded vaccine developers: • To test samples from pre-clinical to Phase II clinical studies for key immunogenicity and efficacy endpoint evaluation • To support SARS-CoV-2 vaccine developers in the pathway towards licensure • To help the identification of Immune Correlates of Protection • To facilitate rapid evaluation, approval, and dissemination of the most effective vaccine candidates

150 Assays available within the Network

Binding antibodies Neutralizing antibodies T cells

Pseudo typed virus Wild type virus ELISA ELISPOT neutralization neutralization • Pseudo particles with • Colorimetric • Peptide pool of the • Stabilized pre- fusion full VSV backbone microneutralization whole S protein length S, RBD, N • Safer testing alternative assay • Cytokines: IFNy (Th1), IL- • Total IgG in serum (no BSL3 required) • Victoria virus isolate 5 (Th2) Planned in December Qualification (Nexelis/PHE) Completed Completed Completed 2020 Planned in December Planned in December Planned in December Validation (Nexelis/PHE) NA 2020 2020 2020 Verification In progress; to be completed In progress; to be completed In progress; to be completed Planned in January 2020 (receiving labs) by the end of 2020 by the end of 2020 by the end of 2020 Average throughput at 480 240 150 (TBD) Nexelis/PHE 450 (per analyst) (per analyst) (per analyst) (samples per week)

• Common key reagents are provided to all the Labs in the Network • Scalable throughput 151 Qualification vs Validation

• The assays transferred are qualified by the transferring lab (Nexelis/PHE) • The receiving labs will perform a verification (tech transfer process) • The same assays will be validated in parallel by the transferring lab; validation plans have been summitted to FDA for feedback

Tech transfer/Verification

Qualification

Validation

152 Apply today for sample testing at the CEPI Centralized Laboratory Network

• All COVID-19 vaccine developers are invited to apply to use the Centralized Laboratory Network up to clinical Phase IIa

• CEPI is working towards expanding the testing to late phase clinical trials

• Bridging with similar efforts (e.g.: the Operation Warp Speed)

• To apply for sample testing, please complete and submit the Sample Analysis Request Form

Any further question? Reach out to [email protected]

153 CEPI Centralized Laboratory Network Why Who When

• To facilitate rapid Qualified assays: The testing service is free evaluation, approval, All COVID-19 vaccine Nexelis (Canada) and dissemination of • Full length S,RBD, N of charge, except for the developers are invited to PHE Portnon Down(UK) the most effective ELISA From October 2020 shipment costs vaccine candidates apply to use the Network Q2 Solutions (US) • Pseudo virus and wild for samples from NIBSC (UK) Press release To check your eligibility, • To standardize type virus immunological testing neutralization assays preclinical up to clinical VisMederi Srl (Italy) complete the Sample of Covid-19 vaccines Phase IIa studies Viroclinics (The Netherlands) Analysis Request Form • IFN-y, IL-5 ELISPOT icddr,b (Bangladesh) THSTI (India)

What Where How

154 For more information: [email protected] Thank you for your attention

www.cepi.net Statistical Approaches for Assessment of Immune Correlates of Protection Peter Gilbert, PhD Professor, Vaccine and Infectious Disease Division (Fred Hutchinson Cancer Research Center)

156 Statistical Approaches for Assessment of Immune Correlates of Protection

COVAX SWAT Workshop COVID-19 Correlates of Protection

Peter Gilbert Statistical Center for the HVTN and the Coronavirus Prevention Network Fred Hutchinson Cancer Research Center Department of Biostatistics, University of Washington November 19, 2020 Outline

• A general approach for immune correlates assessment in phase 3 vaccine efficacy trials • Special considerations for the current context of two vaccines with vaccine efficacy > 90% and few vaccine breakthrough cases

2 COVID Prevention Network (CoVPN)

•Co-conducts with companies USG-sponsored COVID-19 vaccine and mAb trials •Vaccine Trials Leadership: Larry Corey (FHCRC) and Kathleen Neuzil (U Maryland) Laboratory Leadership • Julie McElrath Rafi Ahmed David Montefiori Georgia Tomaras Ralph Baric Mark Dennison Tim Sheahan

CoVPN Statistical Group NIAID Biostatistics Fred Hutch and UW Biostatistics, Colleagues at other departments (e.g., UW Statistics, Emory Biostatistics)

Statistical Leadership Dean Follmann (NIAD) Yonghong Gao (BARDA) Peter Gilbert (FHCRC, UW)

Corey, Mascola, Fauci, Collins. Science (2020) CoVPN Biostatistics Development of an Immune Correlates SAP and Open-Source R Package • Developing a Statistical Analysis Plan for immune correlates assessment for a typical phase 3 trial, publicly posted at Figshare (suitable for version-controlled updates) https://figshare.com/articles/online_resource/CoVPN_COVID- 19_Vaccine_Efficacy_Trial_Immune_Correlates_SAP/13198595 • SAP implemented on Github with application to simulated COVID-19 VE trial data sets • COVID Immune Correlates R package under development • Principle: Open sharing, global community of biostatisticians

Youyi Fong Objectives of Immune Correlates Analyses: Generate Evidence Needed to Qualify an Antibody Biomarker for Use in Vaccine Approval

1. Traditional Approval (based on a ‘validated surrogate’*) Biomarker scientifically well established to reliably predict vaccine efficacy

2. Accelerated Approval (based on a ‘non-validated surrogate’*) Biomarker reasonably likely to predict vaccine efficacy but not yet scientifically well established

o Bridging applications: ▪ Extend the indication of an approved vaccine for broader populations ▪ Rapidly approve or provisionally approve a new vaccine in the same class of vaccines *Nomenclature of Fleming and Powers (2012, Stat Med) Sources of Evidence for Immune Correlates

• Validation of assays and • Phase 3 vaccine efficacy trials biomarkers that: • Longitudinal re-infection studies • Determine overall vaccine efficacy with reasonable precision • Vaccine challenge studies • Store samples post vaccination • NHPs, hamsters from most participants (for • Controlled Human Infection Model subsequent immunologic (CHIM) measurements) • Passive mAb transfer studies • Challenge studies in animals and/or CHIM • Post-exposure prophylaxis • Phase 3 efficacy studies

6 Study Immune Correlates Against Each Study Endpoint Evaluated in Phase 3 Vaccine Efficacy Trials

Clinical Endpoints

Primary Endpoint

8 Biomarker endpoint

More detailed clinical endpoints defined SARS-CoV-2 viral load by post COVID diagnosis follow-up (e.g., mid-turbinate nasal swabs ) 8 Typical OWS Phase 3 Trial Schema

Study Arm Sample Size Immunization Schedule Vaccine 20,000 Day 1 Day 29 Placebo 20,000 Day 1 Day 29

Key time point for an antibody marker surrogate endpoint Main Immunoassays for Immune Correlates Analyses*

2. Pseudovirus 3. Live Virus 1. Binding Ab neutralizing Ab neutralizing Ab Reporter virus Pseudovirus MSD ECL S-2P, RBD and N Full-length molecular clone, Vero-E6 Spike pseudovirus - Lentiviral vector, ffLuc Multiplex Assay cells, nanoLuc reporter (meso scale discovery - reporter Wild-type virus electrochemiluminescence assay) WA isolate, passage 3, Vero-E6 cells Data on Day 57 Ab distributions from a simulated COVID-19 data set

*OWS/CoVPN lab leadership includes Rick Koup, Ruben Donis, Adrian McDermott, Julie McElrath, Rafi Ahmed, David Montefiori, Georgia Tomaras, Ralph Baric, Mark Dennison, Tim Sheahan

10 Structuring Immune Correlates Analyses at Two Levels*

Association • Correlates of Risk (CoRs) Parameters o Correlation of immunologic biomarker in vaccine recipients with outcome (risk prediction)

Causal Effect • Correlates of Protection (CoPs) Parameters o Answer questions directly interpretable in terms of reliability that an immunologic biomarker can be used to predict vaccine efficacy against a study endpoint

Pros Cons CoRs Can obtain definitive answers from A CoR may fail to be a CoP, e.g., due to: the phase 3 data sets 1. Confounding 2. Non-transitivity of vaccine effects on the marker and on the outcome 3. Off-target effects CoPs Results directly interpretable in terms Inferences rely on causal assumptions. needed for qualification for approval Compelling evidence may require multiple decisions (“predict vaccine efficacy”) phase 3 trials + external evidence. *Qin, Gilbert, Corey, McElrath, Self (2007, JID); Plotkin and Gilbert (2012, CID) Objectives of Immune Correlates Analyses of a Phase 3 Trial Data Set: Peak Antibody Correlates

• To assess Day 57 antibody biomarkers as various types of correlates: Association 1. Correlates of risk in vaccine recipients (prediction of outcome) Parameters CoR ▪ Relative and absolute risk of outcome by biomarker level ▪ Machine learning models for predicting outcome from multiple biomarkers / readouts

Causal Effect 2. Controlled VE Parameters ▪ Compare risk under assignment to (vaccine, biomarker value) vs. under assignment to placebo 3. Stochastic interventional VE CoP ▪ Compare risk under assignment to (vaccine, stochastic shift in biomarker distribution) vs. under assignment to placebo 4. Mediators of VE (mechanisms of protection / natural direct and indirect effects) ▪ Proportion of VE mediated by the biomarker 5. Correlates of VE (effect modification / principal stratification) ▪ VE across subgroups defined by biomarker level in vaccine recipients Objectives of Immune Correlates Analyses of a Phase 3 Trial Data Set: Outcome Proximal Correlates

• To assess antibody biomarkers over time, inferred near the times of endpoint occurrence, as various types of immune correlates: 6. Correlates of instantaneous risk in vaccine recipients (e.g., Fu and Gilbert, 2016) CoR ▪ Hazard ratio of outcome across levels of the current value of the biomarker 7. Controlled effects on risk in vaccine recipients CoP ▪ Hazard ratio of outcome if all vaccine recipients are assigned to a fixed marker value profile 8. Mediators of VE (mechanisms of protection) (e.g., Zheng and van der Laan, 2017) ▪ Proportion of VE mediated by the longitudinal biomarker(s) profile

= Blood storage for retrospective virus detection, antibody detection 15 (seroconversion), immunogenicity analyses, and immune correlates analyses One More CoP Approach: Meta-Analysis Evaluation of Surrogate Endpoints from a Set of Phase 3 Trials* Causal Effect Parameters

• Study the association of vaccine effects on the biomarker and on the clinical endpoint IgG IgA • Meta-analysis provides the Illustration with greatest empirical rigor for Rotavirus Vaccines validating a surrogate endpoint Scale) Gabriel, Daniels, Halloran • Need standardized trials (2016, Biometrics) (endpoints, labs / assays, “Comparing Biomarkers

SAPs) as Trial-Level General (Transformed • Assess within-vaccine class Surrogates” surrogate endpoints and

whether biomarkers perform Efficacy Application to rotavirus similarly across multiple VE trials

vaccine platforms Vaccine Vaccine • Special role: Only approach that can assess T cell Vaccine Effect on Immunologic Marker markers as CoPs *e.g., surrogate endpoint evaluation literature from Molenberghs, Buyse, et al. Case-Cohort Two-Phase Sampling Design*: Measure Immunologic Markers in Random Subcohort + All Infection Cases

Design used in many previous VE trial immune correlates analyses

*Prentice RL (1986, Biometrika) OWS Random Subcohort for Measuring Immunologic Biomarkers (N=1620 Participants)

Random Subcohort Sample Sizes for Biomarker Measurement Baseline SARS-CoV-2 Negative2 Baseline SARS-CoV-2 Positive3 (Typical Primary Analysis Cohort) Demographic 1 2 3 4 5 6 1 2 3 4 5 6 Covariate Strata1 Vaccine 150 150 150 150 150 150 50 50 50 50 50 50 Placebo 20 20 20 20 20 20 50 50 50 50 50 50

1Key baseline factors defined by age, high-risk conditions, underrepresented minority status 2CoR analysis focuses on baseline negative vaccine recipients 3Study differences in natural-infection-elicited vs. natural-infection+vaccine-elicited responses

19 Illustration of CoR Analysis Results in Vaccine Recipients (Simulated COVID-19 Data Set: N=20,000 Vaccinees, 63 COVID Endpoints)

Binding Ab to RBD Pseudo Neutralizing AbTiter Estimate absolute risk conditional on the Ab marker and adjusting for covariates

Parameter EW[Y|S=s, A=1, W]

COVID COVID of e.g., 2-phase sampling Cox regression, survey of

R package

Probability Probability Probability

D57 Anti-RBD IgG (EU/ml) (log scale) D57 Pseudo nAb ID80 (log scale) Marker = s subgroups Marker = s subgroups

Binding Ab to RBD Pseudo Neutralizing AbTiter Search for antibody marker thresholds of very low or zero risk (absolute correlate)

Parameter EW[Y|S>=s, A=1, W] e.g., TMLE nonparametric regression using a superlearner of glmnets

Y = Indicator of COVID by 4 months; S = Day 57 Abmarker; A = randomization arm; W = baseline prognosticfactors D57 Anti-RBD IgG (EU/ml) (log scale) D57 Pseudo nAb ID80 (log scale) Marker >= s subgroups Marker >= s subgroups Assessment of Co-Correlates: Illustration of Unbiased Statistical Learning Analysis of Different Sets of Immunologic Markers* Some prediction Month 7 CV-AUC Marker Set [95% CI] beyond chance Application to COVID Trials: Input Variable Sets Baseline risk factors (Base) Base + bAb anti-Spike Base + bAb anti-RBD Base + Pseudo neut (ID50, ID80) Base + Live virus neut (ID50, ID80) Base + bAb + Pseudo neut 11/18/2020 21 Base + bAb + Live neut Base + Pseudo neut + Live neut Base + bAb + Pseudo neut + Liveneut

*Superlearner analysis of the HVTN 505 HIV-1 vaccine efficacy trial (Neidich et al., 2019, JCI)

CV-AUC = Cross-validated area under the ROC curve. All models include baseline variables (Age, BMI, baseline risk score). Fx Ab = ADCP, FcR2a, FcR3a. Data generated by Julie McElrath and Georgia Tomaras labs. Correlate of Risk CoP: Controlled VE (CoR) COVID Phenotypic risk SieveAnalysis CoP: Correlate of VE

Phenotypic Distance ofAcquired SARS-CoV-2 Virus to VaccineStrain Synthesis of Phase 3 Immune Correlates Analyses for multiple endpoints CoP: Mediator of VE 80 AA Sequence + 70 SieveAnalysis 60 External studies 50 (Natural history re-infection; VE 40 30 Challenge studies; 20 Passive Ab transfer studies) 10 = 0 AA Sequence Distance ofAcquired Case Supporting Surrogate VE Overall VE Not VE Through SARS-CoV-2 Virus to VaccineStrain Endpoint Acceptance Through Marker Marker CoP: Stochastic Interventional Effect on Risk CoP: Meta- Baseline Correlate Analysis of VE Correlate of VE Mean Counter- factual COVID risk

Host Factors (Demographics, Prior Exposure) Shift in Adaptive IR Marker 26 Summary and Conclusions (1)

• Standardization of labs/assays/SAPs is needed for interpreting results across phase 3 trials and for meta-analysis • Phase 3 trials need to collect baseline prognostic factors as CoR and CoP analyses should adjust for potential confounders • Nonparametric CoR analyses that integrate machine learning provides robust answers (e.g., threshold searching) • A synthesis of multiple types of CoP analyses, with distinct interpretable outputs, can provide a more complete understanding of CoPs • CoP analyses need sensitivity analyses to understand robustness to departures from causal assumptions • Meta-analysis CoP evaluation will be needed • Opportunity for open process for immune correlates assessment for the global community of biostatisticians

27 Outline

28 Requirements to Learn About a CoP in a Phase 3 Trial

1. Validated immunoassay / immunologic biomarker 2. Demonstrated vaccine efficacy 3. Samples available post vaccination from most participants 4. Ample variability of the immunologic biomarker across vaccine recipients 5. Enough vaccine breakthrough endpoint cases

29 How to Plan Learning About Immune Correlates of Protection Post mRNA Vaccine Efficacy Announcements?

Numbers of Primary Endpoints (Symptomatic COVID Infections) Pfizer Moderna Pooled Placebo Vaccine Placebo Vaccine Placebo Vaccine November 16 90 4 90 5 180 9 Project ~6 weeks later 180 8 180 10 360 18 Project ~6 weeks later 180 20 180 25 360 45 with waning

• What can we learn with < 20 breakthrough cases and VE > 90-95%? • What is the minimum number of vaccine breakthrough cases to support different kinds of immune correlates analyses?

30 Potential Explanations for Rare Vaccine Failure Interrogable Through CoP Analysis

Example of a strong CoP 1. Non-take of the vaccine (may be definable by a variety of assays − a non-mechanistic CoP) 2. Lack of a mechanistic CoP response (e.g., nAb) 3. Antigenic mismatch of exposing virus

Vaccine Recipients

31 If Antigenic-Mismatch Causes Rare Vaccine Failure, Include Antibody Response to Breakthrough Virus in CoP Analyses

Hypothetical Sieve Effect with Pfizer+Moderna Case Counts: VE(match) = 99%; VE(mismatch) = 25% Antigenic Mismatch Cases

200 180 160 140 120

Endpoints 100 80 60

40 Number of 20 0 Vaccine Placebo Mismatch 7 21 Match 2 159

Match Mismatch How Many Vaccine Breakthrough Cases Before Conducting Correlates Analysis?

33 More Vaccine Breakthrough Cases Improves Power to Detect a Correlate of Risk

Relative risk of COVID by Day 57 Ab marker level, for the strength of association detectable with 80% power*

Relative Risk per 10-Fold Increase in Ab *seqDesign R Marker Detectable with 80% Power package, Juraska et al., 2-phase N = 20 0.191 sampling Breslow- N = 25 0.261 Holubkov (1997) logistic regression N = 30 0.309

N = 40 0.377 34 Strongly Parametrized Models Can Characterize a Correlate of Risk (CoR) with Small Numbers of Breakthrough Vaccine Cases*

*Analysis of a simulated COVID-19 VE trial data set 35 Precedent: Found Much CoR Signal Based on Multiple Assays with 25 Vaccine Breakthrough Cases* Some prediction Month 7 CV-AUC Marker Set [95% CI] beyond chance

11/18/2020 36

*Analysis of the HVTN 505 HIV-1 vaccine efficacy trial (Neidich et al., 2019, JCI); Janes et al. (2018, JID), Fong et al. (2019, JID)

CV-AUC = Cross-validated area under the ROC curve. All models include baseline variables (Age, BMI, baseline risk score). Fx Ab = ADCP, FcR2a, FcR3a. Data generated by Julie McElrath and Georgia Tomaras labs. Power for Immune Correlates Analysis Increases with Number of Vaccine Breakthrough Cases

• Value in combining data across the Pfizer and Moderna trials to improve correlates assessment

37 How Many Vaccine Breakthrough Endpoints Will Be Available for Correlates Analyses?

• Expect >=18 symptomatic infections in Pfizer+Moderna trials by January 1, greater if some waning of vaccine efficacy o Underscores the value in combining data from the two trials • These trials test Day 209 samples for seroconversion, to assess VE against asymptomatic infection o Number of vaccine breakthrough asymptomatic infection cases? o Answers may be available ~April 2021

• Endpoints from other phase 3 trials?

38 How to Learn About Immune Correlates from Active Control Clinical Non-Inferiority Studies?

• Even with 90% efficacy of each vaccine, future trials could observe > 90 breakthrough symptomatic infection cases o At 2% background incidence: ▪ N=90,000 vaccinees: in 6 months ▪ N=30,000 vaccinees: in 18 months • Correlates methods that focus on vaccinated individuals, without a placebo arm, may be especially important

39 Summary and Conclusions (2)

• Can learn something useful about CoRs/CoPs with only 15 breakthrough vaccine cases • Yet need to have greater numbers for achieving convincing precision o project 25-50+ vaccine breakthrough cases pooled across Pfizer + Moderna trials over the next 2-3 months o If placebo recipients are crossed over to the vaccine arm, they can be followed to capture additional cases • Active comparator arm trials pre- and post-approval would provide data for additional vaccine breakthrough cases

40 Thank You!

Acknowledgments: Statisticians Co-Working on Preparing for the Immune Correlates Analyses

• David Benkeser • Youyi Fong • Holly Janes • Bhavesh Borate • Peter Gilbert • Michal Juraska • Marco Carone • Nima Hejazi • Kendrick Li • Ivan Diaz • Ying Huang • Alex Luedtke • Mike Fay • Yunda Huang • Lars van der Laan • Dean Follmann • Ollivier Hyrien • Wenbo Zhang

41 Panel Discussion Moderated by Peter Dull Deputy Director, Integrated Clinical Vaccine Development (Bill & Melinda Gates Foundation)

190 Discussion Panel Members and Potential Topics

Panel Members Potential Topics

• Kathleen Neuzil, MD, MPH, FIDSA, Professor in Vaccinology Clinical / Laboratory / Statistical (University of Maryland School of Medicine), Co-Principal Investigator in Vaccines, the COVID-19 Prevention Trials Network ▪ Clinical trial design considerations in the presence of an acceptable immune correlate • Ana Maria Henao-Restrepo, MD, Medical Officer, R&D Blueprint (WHO) ▪ What have we learned about the added value of wild-type neutralizing antibodies over • Jeff Roberts, MD, Associate Director, Medical Countermeasures and pseudoneutralization or binding antibodies? Scientific Affairs (Office of Vaccines, CBER/FDA) ▪ Barriers of use of ECBS-endorsed • Marco Cavaleri, Head of Office, Anti-infectives and Vaccines in the International Standard for anti-SARS-CoV-2 Human Medicines Evaluation Division (EMA) antibody, for new and ongoing studies

▪ How quickly can correlates analyses be made available? Are pooling options across similar Prior speakers also available for questions & discussion: vaccine platforms credible? • Daniel Brasseur, MD, PhD, Former EMA Expert Regulatory • Richard Koup, MD, Deputy Director, Vaccine Research Center ▪ Comment on use of Accelerated (US) or (NIH/NIAID) Conditional (EU) approaches to licensure and how these approaches may be applied to Covid- • Valentina Bernasconi, PhD, Scientist, Preclinical and Immunology (CEPI) 19 vaccines?

• Peter Gilbert, PhD, Professor, Vaccine and Infectious Disease Division ▪ What are steps required to gain regulatory (Fred Hutchinson Cancer Research Center) endorsement for an immune correlate by a sponsor? For a vaccine platform? Privileged and confidential 191 Wrap Up & Next Steps Jakob Cramer Head of Clinical Development (CEPI)

Paul Kristiansen Head of Standards and Assays, Preclinical and Immunology (CEPI)

192 Closing remarks

• Thank you all for your participation and engagement today

• The COVAX Clinical Dev & Ops and Enabling Sciences SWAT Teams plan to continue sharing learnings across developers as we pursue our common goal – a global supply of safe and effective vaccines

• The Clinical Dev & Ops SWAT Team plans to host two additional workshops in December:

• December 17, TBC: Pre-/post-licensure assessments of transmission

• Mid-December, date TBC: Maternal immunization

• The Enabling Sciences SWAT Team will continue to develop and make available tools for the whole vaccine development community

• We will continue to share resources at the website here (https://epi.tghn.org/covax-overview/) and we will also distribute a workshop report to summarize today’s conversation

Privileged and confidential 193 Clinical Development & Operations and Enabling Sciences SWAT Teams

194