This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
WHO COVID-19 Vaccines Research
Will emerging data allow increased reliance on vaccine immune responses for public health and regulatory decision-making? 3 September 2021, virtual consultation Geneva, Switzerland
This is a draft. The content of this document is not final, and the text may be subject to revisions before
publication. The document may not be reviewed, abstracted, quoted, reproduced, transmitted, distributed,
translated or adapted, in part or in whole, in any form or by any means without the permission of the World Health
Organization.
The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed
or recommended by WHO in preference to others of a similar nature that are not mentioned. Errors and omissions
excepted, the names of proprietary products are distinguished by initial capital letters.
2
WHO COVID-19 vaccines research Will emerging data allow increased reliance on vaccine immune responses for
public health and regulatory decision-making?
September 3, 2021
Table of Contents TABLE OF CONTENTS .............................................................................................................................................2
RCTS REPORTING CLINICAL ENDPOINTS FOR VARIANTS ...............................................................................................6 OBSERVATIONAL STUDIES REPORTING CLINICAL ENDPOINTS FOR THE DELTA VARIANT ...................................................6 EFFICACY AND EFFECTIVENESS AGAINST VARIANTS .....................................................................................................7 OVERVIEW OF HUMAN CHALLENGE STUDIES ................................................................................................................8 STUDIES GENERATING BOTH EVIDENCE ON BREAKTHROUGH INFECTIONS AND IMMUNOGENICITY ....................................8 ASSAYS DEVELOPMENT .......................................................................................................................................... 10 OVERVIEW OF ANIMAL STUDIES REPORTING ON IMMUNOLOGICAL ENDPOINTS ............................................................. 10 OVERVIEW OF STUDIES REPORTING T-CELL RESPONSES .......................................................................................... 11 AVAILABLE DATA AND PLANS TO GENERATE ADDITIONAL DATA ................................................................................... 11 DOES EMERGING DATA ALLOW INCREASED RELIANCE ON VACCINE IMMUNE RESPONSES FOR POLICY DECISION-MAKING? ............................................................................................................................................................................. 12
SESSION 2. CURRENT PROPOSALS FOR REVIEW AND SYNTHESIS OF THE EVIDENCE .......................... 13
INITIATIVES TO ANALYZE AND SYNTHESIZE AVAILABLE DATA ...................................................................................... 13 POTENTIAL FOR AUGMENTING INFORMATION ABOUT CORRELATES OF PROTECTION AND RELIABLY ESTABLISHING
CORRELATES OF PROTECTION ................................................................................................................................. 14 DOES EMERGING DATA ALLOW INCREASED RELIANCE ON VACCINE IMMUNE RESPONSES FOR REGULATORY DECISION-MAKING?................................................................................................................................................................ 15 WHAT IS THE RESEARCH AGENDA MOVING FORWARD? ............................................................................................. 16
WHO COVID-19 vaccines research Will emerging data allow increased reliance on vaccine immune responses for
public health and regulatory decision-making?
September 3, 2021
What is the expected role of correlates of protection in this pandemic? A CoP is an immune response that is statistically associated with protection against infection or
disease with a specific pathogen. Different types of CoPs include an absolute correlate
(specific level or threshold of response highly correlated with protection), relative correlate
(level of response variably correlated with protection), and co-correlate (one of two or more
factors that correlate with protection in alternative, additive, or synergistic ways). CoP can be
mechanistic in that the immune response is responsible for protection or non-mechanistic
(formerly called surrogate) in that the immune response substitutes for the true immunologic
CoP which may be unknown or not easily measurable. CoPs are determined through various
means: levels of passively administered or maternal antibody that protect, analysis of immune
responses in protected and unprotected subjects in efficacy trials, observations made on
vaccine failures or immunosuppressed individuals, in human challenge studies, or extrapolated
from animal challenge studies.
A CoP could potentially enable understanding of the likely efficacy of a new COVID vaccine,
the risk of COVID disease for a vaccinee, the risk of disease in a vaccinated population, lot-to-
lot consistency, and could support the licensure or authorization of a vaccine if efficacy trials
become infeasible. There are many potential protective adaptive immune mechanisms
induced by vaccination that could be considered for evaluation as CoPs – various serum
antibody subclasses, mucosal antibodies, CD4+ T cell and CD8+ T cells. The more we learn
about the immune system the more possibilities there are.
Ten principles were outlined regarding CoPs:
1. Must define – protection against what? Infection? Disease? There may be different
CoPs for different clinical endpoints. An example was given for polio.
2. The mechanism of protection against infection is not the same as the mechanisms of
recovery from infection.
3. A large challenge dose can overcome immunity – which is a possibility with the Delta
variant.
4. Most current vaccines protect through antibodies.
5. Correlates may be relative with no sharp distinction between immunity at different
levels; e.g., high titers may protect against infection while lower titers will also protect in
some.
6. Functional antibodies may be better correlates of protection than non-functional
antibodies; for example, in meningococcal infection protective bactericidal antibodies
are needed
7. More than one factor may protect as co-correlates, for example in influenza both
antibodies and cellular immune factors are important for protection.
8. Memory may be a mechanistic CoP, for example Hepatitis B antibodies fade over time
but memory responses prevent disease if a vaccinated individual is exposed.
9. T cell responses may also be important correlates, but these have not been well
defined.
10. Non-mechanistic CoP are useful, for example zoster vaccine is highly protective based
on T cells but antibodies can be measured as a proxy for protection.
6
WHO COVID-19 vaccines research Will emerging data allow increased reliance on vaccine immune responses for
public health and regulatory decision-making?
September 3, 2021
The immune system is redundant, and we frequently have situations where more than one
type of response correlates with protection and multiple markers can be measured. Current
studies stress the importance of antibodies, and we will likely hear modifications as new
evidence emerges due to the complexity of the immune system.
Session 1. Emerging Evidence RCTs reporting clinical endpoints for variants The COVID-NMA (https://covid-nma.com) is an international research initiative supported by
WHO and Cochrane to provide a live mapping of COVID-19 trials. As part of the COVID-NMA
living systematic review, an update of all randomized evidence on efficacy against variants of
concern (VOC) was provided. As of September 2021, 323 registered randomized controlled
clinical trials (RCTs) have assessed COVID-19 vaccine efficacy, with 60 RCTs published. Of
these 7 RCTs report data on VOCs. Two RCTS are available for Alpha, both against
symptomatic confirmed COVID-19 and 3.5-4 months of follow-up. Novavax showed 86.3%
efficacy and AstraZeneca showed 70.4% efficacy. There was some concern on risk of bias in
both studies because they were post-hoc analyses which led to missing data on sequences.
Four RCTs are available for Beta, all against symptomatic confirmed COVID-19 with follow-up
ranging from 1 month to 4 months. Vaccine efficacy (VE) ranged from 10% for the
AstraZeneca vaccine to 100% for the Pfizer/BioNTech vaccine. Similar to Alpha, all had some
concern for risk of bias due to post-hoc analyses. The AstraZeneca study in particular had
large confidence intervals due to a small sample size. One RCT is available for Delta and the
Bharat Biotech vaccine, with symptomatic confirmed COVID-19 as the outcome with 3.3
months of follow-up. VE was 65.2% with some risk of bias because it was a post-hoc analysis.
Overall, there is limited evidence from RCTs on VOCs and consistent methodological issues
including post-hoc analyses that lead to lack of power, missing outcome data due to lack of
sequencing and, heterogeneous measurements of outcomes (both direct and indirect
evidence). Continued observational studies are needed to provide further evidence.
Observational studies reporting clinical endpoints for the delta variant As part of COVID-NMA, 127 observational studies on VOC have been identified, 100 have
relevant outcomes to assess VE and 28 are studies on the Delta variant. Of those, 13 meet the
strict eligibility criteria including using a control/non-vaccinated group to estimate vaccine
effectiveness and with some attempt to control confounding. These studies are cohort and
test-negative case-control studies conducted in the general public and in healthcare workers
taking place in India, Canada, the USA, UK, Qatar, Scotland, and Israel. After two doses, VE
against Delta SARS-CoV-2 infection (asymptomatic or symptomatic) was about 60-86%, with
no meaningful differences between vaccines (AstraZeneca, Pfizer, Moderna, Coronavac). As
expected, after only one dose, VE was lower than after two doses. After two doses, VE against
FDA has reviewed the validation of both assays which use similar technologies (lentivirus
backbone, full-length spike, luminescence, and transfection) but generate a 3-fold difference
in titers for convalescent and vaccine sera. This may be due to differences in PsV production –
Duke uses cells expressing TMPRSS2. Calibration is needed to support decision-making based
on data from these two assays including identifying immune correlates, licensure,
immunobridging regulatory approvals, and boosting decisions.
Three sample sets were used including convalescent sera from early in the pandemic, samples
from recipients of the Moderna vaccine, and the WHO International Standard (WHO-IS). Three
calibration approaches were used; 1) calibrate to WHO-IS, 2) calibrate to the convalescent
sera using bivariate normal distribution, and 3) calibrate to convalescent sera using linear
regression. As expected, the ID50 titers were consistently 3-fold higher in the Monogram assay
compared to Duke. All three calibration approaches worked well to bring these into
equivalence with the greatest concordance correlation coefficient using the second method
(CCC: 0.87), though the WHO-IS also performed well (CCC: 0.75). Comparisons were made
using the arithmetic mean, geometric mean and median and the arithmetic mean performed
the best.
Overview of animal studies reporting on immunological endpoints The first non-human primate (NHP) studies conducted in 2020 showed that vaccines led to
protection at a reasonably high level. Initial correlates analyses demonstrated nAbs and bAbs
correlated inversely with peak viral loads in nasal swabs and bronchoalveolar lavage. In an
IgG adoptive transfer study examining the mechanisms of protection, purified IgG from
convalescent macaques was given to naïve animals to determine if this protected animals,
and if so to define a NAb threshold for protection. A dose-dependent response was observed
where the highest dose completely protected animals, the lowest didn’t protect but showed
lowered peaks and faster resolution of virus. This showed that antibodies, even isolated from
other immune components, could provide protection. A CD8 T cell depletion study was also
conducted to determine the role of CD8 T-cell-dependent responses in protection again re-
challenge. CD8 T cell depletion reduced protection against re-challenge in those subjects
with waning Ab titers. This data suggested that Abs alone can protect, but also that cellular
immune responses contribute to protection when Ab titers are borderline or sub-protective.
The levels seen in animals are not too different than what is seen as protective levels in humans
with the Moderna and Oxford studies.
To date, all data has been based on the ancestral sequence and these early studies may not
apply with VOCs. Studies show reduced responses to Beta and Delta for Pfizer, Moderna, and
J&J vaccines. bAb and nAb protective thresholds will likely be different for variants, especially
those with increased transmissibility such as Delta. Antibody responses do not appear to fully
WHO COVID-19 vaccines research Will emerging data allow increased reliance on vaccine immune responses for
public health and regulatory decision-making?
September 3, 2021
neutralization for variants in the future. Experts noted that without standardization across
cellular immune studies, comparison and integration of results from different studies are
difficult. Animal models could be used to move the vaccine development pipeline forward
and answer some of these remaining questions.
There was an important reminder that policy decision-making depends on the context - with
LMICs focusing on approval towards vaccines preventing severe disease and high-income
countries also considering how vaccines may prevent infection and transmission, though data
on these outcomes are limited. These endpoints are likely to have different CoPs. There was a
recommendation to have an immunobridging toolkit where developers should consistently
provide a packet of information to see the breadth of immune response produced – across
VOCs, using WHO reagent panels of sera, standardized pseudoviruses, standard proteins for
binding assays, and reagent toolkits to better compare vaccines. In general, the community
needs to be flexible and not focus only on nAb but look at a broad range of responses to mine
for mechanistic correlates of immunity.
Session 2. Current proposals for review and synthesis of the evidence Initiatives to analyze and synthesize available data Three presentations were given to describe initiatives to analyze and synthesize available
data.
A model was previously developed to predict protective efficacy in phase 1-3 trials based on
the mean and distribution of nAb titers. This analysis was adapted to predict protection against
VOCs. Neutralizing activity against the ancestral SARS-CoV-2 can be used to predict
neutralization of the VOC, with all examined vaccines showing similar drops in neutralization to
the variants (Alpha 1.6 old, Beta 8.8 fold, Delta 3.9 fold). Adapting the curves based on VOC
resulted in 13/14 studies falling within the 95% CI of the study VE. This same method could be
used to predict efficacy of a new vaccine by estimating neutralization against the ancestral
virus and benchmarking in the same assay to different vaccines to use as marker points. The
same could potentially be done with a new variant to estimate the fold-drop compared to
the ancestral strain to know how much to shift the curve. It was noted that the model
continues to work well despite the increased rapidity of Delta infection compared to other
variants.
The second presentation built on previous work that showed the relationship between efficacy
and nAb or bAb. The first studies had to be expressed as a ratio of GMTs to convalescent
patients, but the current studies standardized results in a convenience sample of SARS-CoV-2
naïve subjects donated as part of national rollouts instead of a clinical trial. Results showed
strong correlations between binding Ab to spike and RBD. Spike IgG binding levels explained
97.4% of the variance in VE. Binding antibodies to Spike and RBD were correlated with nAb. A
population-based approach, as seen with pneumococcal conjugate vaccine (PCV), could
be used to relate the observed distribution of Ab in a subset of the immunized population to