C. Jessica E. Metcalf

Serology in a time of pandemic: consequences for population dynamics

cmetcalf@princeton.edu

Every emergence event has contained surprises

-H1N1 did not attack older individuals-MERS has not gone on to cause widespread outbreaks-Ebola was not thought likely to spread so far or so fast-Zika was not thought to be pathogenic-SARS-CoV-2 case burden (infection?) is low in children

2005 2010 2015

SARS 2003

H1N1 2009

Ebola 2014

MERS2013

Zika2015

2020

Emergence time line

SARS-CoV-2 2019

Metcalf & Lessler 2017 Science

Every emergence event has contained surprises

-H1N1 did not attack older individuals-MERS has not gone on to cause widespread outbreaks-Ebola was not thought likely to spread so far or so fast-Zika was not thought to be pathogenic-SARS-CoV-2 case burden (infection?) is low in children

2005 2010 2015

SARS 2003

H1N1 2009

Ebola 2014

MERS2013

Zika2015

2020

Emergence time line

SARS-CoV-2 2019

Metcalf & Lessler 2017 Science

Aims

1. Describing transmission Early outbreak growth The role of heterogeneities

2. The longer term: How herd immunity emerges

3. Susceptibility as ‘immune dark matter’ Serology for model calibrationSerology to understand immunity

4. Serology to meet future pandemics

5. Conclusions: the use case

Two key quantities: R0, here = 2Serial interval:

initial case

1. Describing transmission

Two key quantities: R0, here = 2Serial interval:

1. Describing transmission

Two key quantities: R0, here = 2Serial interval:

1. Describing transmission

Two key quantities: R0, here = 2Serial interval:

… … … … … … … … …… … … … … …

1. Describing transmission

Two key quantities: R0, here = 2Serial interval:

… … … … … … … … …… … … … … …

With a serial interval of ~ 1 week and an R0 of 2, cases double approximately every week (R0 estimate: ~ 2 - 3)

1. Describing transmission

Two key quantities: R0, here = 2Serial interval:

… … … … … … … … …… … … … … …

https://coronavirus.jhu.edu/map.html

Feb 01 Mar 01 Apr 01

1100

10000

Deaths

Feb 01 Mar 01 Apr 01

1100

10000

Cases

New JerseyCaliforniaWashingtonNew York

Exponential growth

1. Describing transmission

Two key quantities: R0, here = 2Serial interval:

… … … … … … … … …… … … … … …

https://coronavirus.jhu.edu/map.html

Feb 01 Mar 01 Apr 01

1100

10000

Deaths

Feb 01 Mar 01 Apr 01

1100

10000

Cases

New JerseyCaliforniaWashingtonNew York

Exponential growth

Cases: under-report infections. How many people could still be infected?

1. Describing transmission

Not ‘over-dispersed’

60% causes 80%

100% of introductions cause onward transmission

1. Describing transmission: superspreading events

Not ‘over-dispersed’

60% causes 80%

‘Over-dispersed’

20% causes 80%

100% of introductions cause onward transmission

40% of introductions cause onward transmission

Superspreading events: with R0 = 3, on average, 1 person infects 3 others, but most people infect no one, and a few infect very many

outbreaks are more explosive, but also more prone to extinction.

1. Describing transmission: superspreading events

Not ‘over-dispersed’

60% causes 80%

‘Over-dispersed’

20% causes 80%

100% of introductions cause onward transmission

40% of introductions cause onward transmission

https://hopkinsidd.github.io/nCoV-Sandbox/DispersionExploration.html

Map observed (165 introductions; 84 local cases)

Re, # secondary cases per case

prop

ortio

n in

fect

ing

80%

1. Describing transmission: superspreading events

Aims

1. Describing transmission Early outbreak growth The role of heterogeneities

2. The longer term: How herd immunity emerges

3. Susceptibility as ‘immune dark matter’ Serology for model calibrationSerology to understand immunity

4. Serology to meet future pandemics

5. Conclusions: the use case

recovered

infected

‘herd immunity’

out$time

RE

0.5

1.0

1.5

2.0

2.5

3.0

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0

Proportion

Susceptible

Infected

Recovered

R0 = 3

Pro

porti

on

Time (weeks)

susceptible

infected

Metcalf et al. 2015 Trends in Immunology

2. The longer term

recovered

infected

‘herd immunity’

out$time

RE

0.5

1.0

1.5

2.0

2.5

3.0

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0

Proportion

Susceptible

Infected

Recovered

R0 = 3

Pro

porti

on

Time (weeks)

susceptible

infected recovered

Metcalf et al. 2015 Trends in Immunology

2. The longer term

recovered

infected

susceptible

‘herd immunity’

indirectly protected

out$time

RE

0.5

1.0

1.5

2.0

2.5

3.0

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0

Proportion

Susceptible

Infected

Recovered

R0 = 3

Pro

porti

on

recovered

Metcalf et al. 2015 Trends in Immunology

2. The longer term

recovered

infected

susceptible

‘herd immunity’

indirectly protected

out$time

RE

0.5

1.0

1.5

2.0

2.5

3.0

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0

Proportion

Susceptible

Infected

Recovered

R0 = 3

Pro

porti

on

Metcalf et al. 2015 Trends in Immunology

2. The longer term

recovered

infected

susceptible

‘herd immunity’

indirectly protected

out$time

RE

0.5

1.0

1.5

2.0

2.5

3.0

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0

Proportion

Susceptible

Infected

Recovered

R0 = 3

Pro

porti

on

Metcalf et al. 2015 Trends in Immunology

pc = 1 − 1/R0

critical threshold for vaccination

2. The longer term

recovered

infected

susceptible

‘herd immunity’

indirectly protected

out$time

RE

0.5

1.0

1.5

2.0

2.5

3.0

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0

Proportion

Susceptible

Infected

Recovered

R0 = 3

Pro

porti

on

Metcalf et al. 2015 Trends in Immunology

pc = 1 − 1/R0

critical threshold for vaccination

2. The longer term

• 1/R0 is conservative (heterogeneities)

• The downslope is long

Aims

1. Describing transmission Early outbreak growth The role of heterogeneities

2. The longer term: How herd immunity emerges

3. Susceptibility as ‘immune dark matter’ Serology for model calibrationSerology to understand immunity

4. Serology to meet future pandemics

5. Conclusions: the use case

Cases and deaths:

Susceptible

Infected

Recovered

3. Susceptibility as immune ‘dark matter’

Cases and deaths:Provide an incomplete window onto the epidemic

Susceptible

Infected

Recovered

Viewpoint

www.thelancet.com Published online April 5, 2016 http://dx.doi.org/10.1016/S0140-6736(16)30164-7 1

Use of serological surveys to generate key insights into the changing global landscape of infectious diseaseC Jessica E Metcalf, Jeremy Farrar, Felicity T Cutts, Nicole E Basta, Andrea L Graham, Justin Lessler, Neil M Ferguson, Donald S Burke, Bryan T Grenfell

A central conundrum in the study of infectious disease dynamics is to defi ne the landscape of population immunity. The proportion of individuals protected against a specifi c pathogen determines the timing and scale of outbreaks, and the pace of evolution for infections that can evade prevailing humoral immunity. Serological surveys provide the most direct measurement to defi ne the immunity landscape for many infectious diseases, yet this methodology remains underexploited. To address this gap, we propose a World Serology Bank and associated major methodological developments in serological testing, study design, and quantitative analysis, which could drive a step change in our understanding and optimum control of infectious diseases.

Epidemic dynamics result from an interaction between the contagious spread of infection, the resulting depletion of population susceptibility, and its replenishment via births, immigration, or waning immunity. Under-standing this interaction is key to assess the eff ect of vaccination, which artifi cially reduces the number of people susceptible to infection.

Researchers mainly observe infection dynamics and the eff ect of population (or herd) immunity in limiting spread via surveillance of clinically apparent cases of infection or deaths. This method has led to some powerful insights;1 however, even in the simplest instances in which subclinical infection is uncommon, cases only provide information about the dynamics of infection. Susceptibility and immunity are hidden variables. For infections that people can be completely immunised against, such as measles, susceptible reconstruction can be used to estimate immune profi les,2 but infection prevalence and vaccination coverage records are frequently inadequate to capture key social and geographical heterogeneities. Additionally, inference is weakened if the risk of infection is low.

Serological surveys (usually used to quantify the proportion of people positive for a specifi c antibody or, better yet, the titre or concentrations of an antibody) are potentially the most direct and informative technique available to infer the dynamics of a population’s susceptibility and level of immunity. However, the use of current serological tests varies greatly depending on type of pathogens and there are major methodological gaps in some areas for some pathogens and tests. In terms of use of current serological methods, infections can be classifi ed into four broad groups (appendix). The fi rst group contains acute immunising, antigenically stable pathogens (eg, measles, rubella, and smallpox) for which serology provides a strong signal of lifetime

protection and a clear marker of past infection (or vaccination).

The second group contains immunising, but antigenically variable pathogens (eg, infl uenza, invasive bacterial diseases, and dengue). Despite complexities (appendix), serology in some cases can provide powerful evidence, both for vaccine formulation and pandemic planning.3,4 A serum bank would have been extremely useful in interpretation of the unusual profi le of susceptibility associated with age in the 2009 infl uenza pandemic.3 For these fi rst two groups, if suitable serum banks existed, the deployment of current serological tests could have helped to clarify the association between serological profi les and protective immunity.

The third group includes infections for which infection-induced antibodies are not thought to be protective, such as tuberculosis in which the targets of the immune response vary with stage of infection; malaria, whereby infected erythrocytes generate several antibodies whose individual importance has not been fully elucidated (and might indicate exposure rather than protection5); and HIV. Although antibodies might not be representative of immunity against a pathogen, they do show current or previous infection.

Finally, the last broad grouping consists of infections that do not lead to reliably sustained, measurable antibody responses or for which presence of specifi c antibodies do not correlate with protection from future infection. These include many enteric infections and the human papillomavirus. In these cases, serological data can nonetheless be valuable to assess a population’s coverage of vaccine programmes if vaccination leads to long-lasting antibody responses.

In the context of public health, for immunisation against group one infections, vaccination programmes aim to protect vaccinated individuals and indirectly protect unvaccinated individuals by maintaining high population immunity.1 If a valid correlate of immunity is measurable in sera, serological surveys could be used to identify population subgroups in which immunity is low, or even to identify individuals in whom immunity has waned, and directly inform targeted vaccination strategies (appendix).

Household surveys are a major source of data for vaccination coverage in low-income countries.6 The recent extension of eff orts to measure biomarkers for infections such as HIV (eg, the Demographic Health Surveys) could provide infrastructure for sera collection for an expanded range of infections, thus leveraging an existing platform. For many infections, however, to

Published OnlineApril 5, 2016http://dx.doi.org/10.1016/S0140-6736(16)30164-7

Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ, USA (C J E Metcalf PhD, A L Graham PhD, Prof B T Grenfell PhD); Fogarty International Center, National Institute of Health, Bethesda, MD, USA (C J E Metcalf, N E Basta PhD, A L Graham, Prof B T Grenfell); Wellcome Trust, Gibbs Building, London, UK (J Farrar MD); London School of Hygiene & Tropical Medicine, London, UK (Prof F T Cutts MD);

Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis, MN, USA (N E Basta); Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA (J Lessler PhD); Department of Medicine, School of Public Health, Imperial College London, London, UK (Prof N M Ferguson DPhil); and Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA (Prof D S Burke MD)

Correspondence to:Dr C Jessica E Metcalf, Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544, USAcmetcalf@princeton.edu

See Online for appendix

For more on the Demographic Health Surverys see http://www.dhsprogram.com/

3. Susceptibility as immune ‘dark matter’

Cases and deaths:Provide an incomplete window onto the epidemic

Susceptible

Infected

Recovered

Viewpoint

www.thelancet.com Published online April 5, 2016 http://dx.doi.org/10.1016/S0140-6736(16)30164-7 1

Use of serological surveys to generate key insights into the changing global landscape of infectious diseaseC Jessica E Metcalf, Jeremy Farrar, Felicity T Cutts, Nicole E Basta, Andrea L Graham, Justin Lessler, Neil M Ferguson, Donald S Burke, Bryan T Grenfell

A central conundrum in the study of infectious disease dynamics is to defi ne the landscape of population immunity. The proportion of individuals protected against a specifi c pathogen determines the timing and scale of outbreaks, and the pace of evolution for infections that can evade prevailing humoral immunity. Serological surveys provide the most direct measurement to defi ne the immunity landscape for many infectious diseases, yet this methodology remains underexploited. To address this gap, we propose a World Serology Bank and associated major methodological developments in serological testing, study design, and quantitative analysis, which could drive a step change in our understanding and optimum control of infectious diseases.

Epidemic dynamics result from an interaction between the contagious spread of infection, the resulting depletion of population susceptibility, and its replenishment via births, immigration, or waning immunity. Under-standing this interaction is key to assess the eff ect of vaccination, which artifi cially reduces the number of people susceptible to infection.

Researchers mainly observe infection dynamics and the eff ect of population (or herd) immunity in limiting spread via surveillance of clinically apparent cases of infection or deaths. This method has led to some powerful insights;1 however, even in the simplest instances in which subclinical infection is uncommon, cases only provide information about the dynamics of infection. Susceptibility and immunity are hidden variables. For infections that people can be completely immunised against, such as measles, susceptible reconstruction can be used to estimate immune profi les,2 but infection prevalence and vaccination coverage records are frequently inadequate to capture key social and geographical heterogeneities. Additionally, inference is weakened if the risk of infection is low.

Serological surveys (usually used to quantify the proportion of people positive for a specifi c antibody or, better yet, the titre or concentrations of an antibody) are potentially the most direct and informative technique available to infer the dynamics of a population’s susceptibility and level of immunity. However, the use of current serological tests varies greatly depending on type of pathogens and there are major methodological gaps in some areas for some pathogens and tests. In terms of use of current serological methods, infections can be classifi ed into four broad groups (appendix). The fi rst group contains acute immunising, antigenically stable pathogens (eg, measles, rubella, and smallpox) for which serology provides a strong signal of lifetime

protection and a clear marker of past infection (or vaccination).

The second group contains immunising, but antigenically variable pathogens (eg, infl uenza, invasive bacterial diseases, and dengue). Despite complexities (appendix), serology in some cases can provide powerful evidence, both for vaccine formulation and pandemic planning.3,4 A serum bank would have been extremely useful in interpretation of the unusual profi le of susceptibility associated with age in the 2009 infl uenza pandemic.3 For these fi rst two groups, if suitable serum banks existed, the deployment of current serological tests could have helped to clarify the association between serological profi les and protective immunity.

The third group includes infections for which infection-induced antibodies are not thought to be protective, such as tuberculosis in which the targets of the immune response vary with stage of infection; malaria, whereby infected erythrocytes generate several antibodies whose individual importance has not been fully elucidated (and might indicate exposure rather than protection5); and HIV. Although antibodies might not be representative of immunity against a pathogen, they do show current or previous infection.

Finally, the last broad grouping consists of infections that do not lead to reliably sustained, measurable antibody responses or for which presence of specifi c antibodies do not correlate with protection from future infection. These include many enteric infections and the human papillomavirus. In these cases, serological data can nonetheless be valuable to assess a population’s coverage of vaccine programmes if vaccination leads to long-lasting antibody responses.

In the context of public health, for immunisation against group one infections, vaccination programmes aim to protect vaccinated individuals and indirectly protect unvaccinated individuals by maintaining high population immunity.1 If a valid correlate of immunity is measurable in sera, serological surveys could be used to identify population subgroups in which immunity is low, or even to identify individuals in whom immunity has waned, and directly inform targeted vaccination strategies (appendix).

Household surveys are a major source of data for vaccination coverage in low-income countries.6 The recent extension of eff orts to measure biomarkers for infections such as HIV (eg, the Demographic Health Surveys) could provide infrastructure for sera collection for an expanded range of infections, thus leveraging an existing platform. For many infections, however, to

Published OnlineApril 5, 2016http://dx.doi.org/10.1016/S0140-6736(16)30164-7

Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ, USA (C J E Metcalf PhD, A L Graham PhD, Prof B T Grenfell PhD); Fogarty International Center, National Institute of Health, Bethesda, MD, USA (C J E Metcalf, N E Basta PhD, A L Graham, Prof B T Grenfell); Wellcome Trust, Gibbs Building, London, UK (J Farrar MD); London School of Hygiene & Tropical Medicine, London, UK (Prof F T Cutts MD);

Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis, MN, USA (N E Basta); Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA (J Lessler PhD); Department of Medicine, School of Public Health, Imperial College London, London, UK (Prof N M Ferguson DPhil); and Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA (Prof D S Burke MD)

Correspondence to:Dr C Jessica E Metcalf, Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544, USAcmetcalf@princeton.edu

See Online for appendix

For more on the Demographic Health Surverys see http://www.dhsprogram.com/

Serology:Measurement of antibodies can complete the picture

3. Susceptibility as immune ‘dark matter’

0 2 4 6 8 10 12

0.00

0.05

0.10

0.15

0 2 4 6 8 10 12

0.0

0.2

0.4

0.6

0.8

1.0

Pro

porti

on s

usce

ptib

le

infected

3. Susceptibility as immune ‘dark matter’

0 2 4 6 8 10 12

0.00

0.05

0.10

0.15

0 2 4 6 8 10 12

0.0

0.2

0.4

0.6

0.8

1.0

Pro

porti

on s

usce

ptib

le

reported cases (1 in 5)

infected

3. Susceptibility as immune ‘dark matter’

0 2 4 6 8 10 12

0.00

0.05

0.10

0.15

0 2 4 6 8 10 12

0.0

0.2

0.4

0.6

0.8

1.0

Pro

porti

on s

usce

ptib

le

reported cases (1 in 5)

infected susceptible

3. Susceptibility as immune ‘dark matter’

0 2 4 6 8 10 12

0.00

0.05

0.10

0.15

0 2 4 6 8 10 12

0.0

0.2

0.4

0.6

0.8

1.0

Pro

porti

on s

usce

ptib

le

reported cases (1 in 5)

infected susceptible

span data early in the outbreak….

3. Susceptibility as immune ‘dark matter’

0 2 4 6 8 10 12

0.00

0.05

0.10

0.15

observedfittedtrue

0 2 4 6 8 10 12

0.0

0.2

0.4

0.6

0.8

1.0

Pro

porti

on s

usce

ptib

le

fittedtrue

infected

fit 1

susceptible

fit 1

span data

Estimated reporting of 3 in 5 infections, estimated transmission accurate, but duration infection shorter, and later start.

early in the outbreak….

3. Susceptibility as immune ‘dark matter’

span data

fit 1

susceptible

fit 1

fit 2

fit 2

0 2 4 6 8 10 12

0.00

0.05

0.10

0.15

observedfittedtrue

0 2 4 6 8 10 12

0.0

0.2

0.4

0.6

0.8

1.0

Pro

porti

on s

usce

ptib

le

fittedtrue

infected

early in the outbreak….

Estimated reporting of 1 in 10 infections, estimated transmission of 2 x the truth, duration infection longer, and earlier start.

Estimated reporting of 3 in 5 infections, estimated transmission accurate, but duration infection shorter, and later start.

3. Susceptibility as immune ‘dark matter’

span data

fit 1

susceptible

fit 1

fit 2

fit 2

0 2 4 6 8 10 12

0.00

0.05

0.10

0.15

observedfittedtrue

0 2 4 6 8 10 12

0.0

0.2

0.4

0.6

0.8

1.0

Pro

porti

on s

usce

ptib

le

fittedtrue

infected

Susceptibility integrates over the past, and thus provides much more information than cases.

Estimated reporting of 1 in 10 infections, estimated transmission of 2 x the truth, duration infection longer, and earlier start.

Estimated reporting of 3 in 5 infections, estimated transmission accurate, but duration infection shorter, and later start.

https://labmetcalf.shinyapps.io/serol1/

3. Susceptibility as immune ‘dark matter’

Susceptible

Infected

Recovered

https://science.sciencemag.org/content/early/2020/04/14/science.abb5793?rss=1

What will 'seropositive' mean?

‘Recovereds’ may not be susceptible to re-infection for at least a season, based on:

-Experimental evidence (re-infection fails)-Indirect math models (signatures of immunity)-Profile of immunity following COVID-19 infection

3. Susceptibility as immune ‘dark matter’

Susceptible

Infected

Recovered

Testing health care workers, other essential professions could contribute to moving back to normality.

What will 'seropositive' mean?

‘Recovereds’ may not be susceptible to re-infection for at least a season, based on:

-Experimental evidence (re-infection fails)-Indirect math models (signatures of immunity)-Profile of immunity following COVID-19 infection

https://science.sciencemag.org/content/early/2020/04/14/science.abb5793?rss=1

3. Susceptibility as immune ‘dark matter’

Aims

1. Describing transmission Early outbreak growth The role of heterogeneities

2. The longer term: How herd immunity emerges

3. Susceptibility as ‘immune dark matter’ Serology for model calibrationSerology to understand immunity

4. Serology to meet future pandemics

5. Conclusions: the use case

Prognostic serology

Measles:Cross-sectional annual age serology was used to trigger a vaccination campaign in England (Babad et al. 1995)

A convenience sample predicted high outbreak risk in Madagascar (Winter et al. 2018); the outbreak then sadly occurred (2018-2019)

H1N1: “Representative serological surveys stand out as a critical source of data with which to reduce uncertainty around policy choices […]” (Kerkove et al. 2010)

https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1000275

4. Serology to meet future pandemics?

Prognostic serology

Measles:Cross-sectional annual age serology was used to trigger a vaccination campaign in England (Babad et al. 1995)

A convenience sample predicted high outbreak risk in Madagascar (Winter et al. 2018); the outbreak then sadly occurred (2018-2019)

H1N1: “Representative serological surveys stand out as a critical source of data with which to reduce uncertainty around policy choices […]” (Kerkove et al. 2010)

https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1000275

4. Serology to meet future pandemics?

A Global Immunological Observatory

Data MethodsLaboratory techniques Statistical approachesViewpoint

www.thelancet.com Published online April 5, 2016 http://dx.doi.org/10.1016/S0140-6736(16)30164-7 1

Use of serological surveys to generate key insights into the changing global landscape of infectious diseaseC Jessica E Metcalf, Jeremy Farrar, Felicity T Cutts, Nicole E Basta, Andrea L Graham, Justin Lessler, Neil M Ferguson, Donald S Burke, Bryan T Grenfell

A central conundrum in the study of infectious disease dynamics is to defi ne the landscape of population immunity. The proportion of individuals protected against a specifi c pathogen determines the timing and scale of outbreaks, and the pace of evolution for infections that can evade prevailing humoral immunity. Serological surveys provide the most direct measurement to defi ne the immunity landscape for many infectious diseases, yet this methodology remains underexploited. To address this gap, we propose a World Serology Bank and associated major methodological developments in serological testing, study design, and quantitative analysis, which could drive a step change in our understanding and optimum control of infectious diseases.

Epidemic dynamics result from an interaction between the contagious spread of infection, the resulting depletion of population susceptibility, and its replenishment via births, immigration, or waning immunity. Under-standing this interaction is key to assess the eff ect of vaccination, which artifi cially reduces the number of people susceptible to infection.

Researchers mainly observe infection dynamics and the eff ect of population (or herd) immunity in limiting spread via surveillance of clinically apparent cases of infection or deaths. This method has led to some powerful insights;1 however, even in the simplest instances in which subclinical infection is uncommon, cases only provide information about the dynamics of infection. Susceptibility and immunity are hidden variables. For infections that people can be completely immunised against, such as measles, susceptible reconstruction can be used to estimate immune profi les,2 but infection prevalence and vaccination coverage records are frequently inadequate to capture key social and geographical heterogeneities. Additionally, inference is weakened if the risk of infection is low.

Serological surveys (usually used to quantify the proportion of people positive for a specifi c antibody or, better yet, the titre or concentrations of an antibody) are potentially the most direct and informative technique available to infer the dynamics of a population’s susceptibility and level of immunity. However, the use of current serological tests varies greatly depending on type of pathogens and there are major methodological gaps in some areas for some pathogens and tests. In terms of use of current serological methods, infections can be classifi ed into four broad groups (appendix). The fi rst group contains acute immunising, antigenically stable pathogens (eg, measles, rubella, and smallpox) for which serology provides a strong signal of lifetime

protection and a clear marker of past infection (or vaccination).

The second group contains immunising, but antigenically variable pathogens (eg, infl uenza, invasive bacterial diseases, and dengue). Despite complexities (appendix), serology in some cases can provide powerful evidence, both for vaccine formulation and pandemic planning.3,4 A serum bank would have been extremely useful in interpretation of the unusual profi le of susceptibility associated with age in the 2009 infl uenza pandemic.3 For these fi rst two groups, if suitable serum banks existed, the deployment of current serological tests could have helped to clarify the association between serological profi les and protective immunity.

The third group includes infections for which infection-induced antibodies are not thought to be protective, such as tuberculosis in which the targets of the immune response vary with stage of infection; malaria, whereby infected erythrocytes generate several antibodies whose individual importance has not been fully elucidated (and might indicate exposure rather than protection5); and HIV. Although antibodies might not be representative of immunity against a pathogen, they do show current or previous infection.

Finally, the last broad grouping consists of infections that do not lead to reliably sustained, measurable antibody responses or for which presence of specifi c antibodies do not correlate with protection from future infection. These include many enteric infections and the human papillomavirus. In these cases, serological data can nonetheless be valuable to assess a population’s coverage of vaccine programmes if vaccination leads to long-lasting antibody responses.

In the context of public health, for immunisation against group one infections, vaccination programmes aim to protect vaccinated individuals and indirectly protect unvaccinated individuals by maintaining high population immunity.1 If a valid correlate of immunity is measurable in sera, serological surveys could be used to identify population subgroups in which immunity is low, or even to identify individuals in whom immunity has waned, and directly inform targeted vaccination strategies (appendix).

Household surveys are a major source of data for vaccination coverage in low-income countries.6 The recent extension of eff orts to measure biomarkers for infections such as HIV (eg, the Demographic Health Surveys) could provide infrastructure for sera collection for an expanded range of infections, thus leveraging an existing platform. For many infections, however, to

Published OnlineApril 5, 2016http://dx.doi.org/10.1016/S0140-6736(16)30164-7

Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ, USA (C J E Metcalf PhD, A L Graham PhD, Prof B T Grenfell PhD); Fogarty International Center, National Institute of Health, Bethesda, MD, USA (C J E Metcalf, N E Basta PhD, A L Graham, Prof B T Grenfell); Wellcome Trust, Gibbs Building, London, UK (J Farrar MD); London School of Hygiene & Tropical Medicine, London, UK (Prof F T Cutts MD);

Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis, MN, USA (N E Basta); Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA (J Lessler PhD); Department of Medicine, School of Public Health, Imperial College London, London, UK (Prof N M Ferguson DPhil); and Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA (Prof D S Burke MD)

Correspondence to:Dr C Jessica E Metcalf, Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544, USAcmetcalf@princeton.edu

See Online for appendix

For more on the Demographic Health Surverys see http://www.dhsprogram.com/

Convenience samples Longitudinal age serology

4. Serology to meet future pandemics?

Aims

1. Describing transmission Early outbreak growth The role of heterogeneities

2. The longer term: How herd immunity emerges

3. Susceptibility as ‘immune dark matter’ Serology for model calibrationSerology to understand immunity

4. Serology to meet future pandemics

5. Conclusions: the use case

• Proportion seronegative at one time-point tells us ‘how many people are left to infect’: where are we on the epidemic trajectory?

• Longitudinal samples that track individuals’ status through time will help establish: what does seropositivity ‘mean’ for immunity?

• Serology patterns over age and through time will help demystify the role of children in transmission (linked to core question of school closures) and establish risks to older populations via transmission (models + information on contact patterns will be required).

• Larger sampling (including internationally!) should clarify transmission and burden effects (from socio-economic, to co-morbidities, to genetics).

• Serology in tandem with the step down of Non-Pharmaceutical Interventions could establish the costs and benefits of different interventions.

5. Serology: the use case

• Proportion seronegative at one time-point tells us ‘how many people are left to infect’: where are we on the epidemic trajectory?

• Longitudinal samples that track individuals’ status through time will help establish: what does seropositivity ‘mean’ for immunity?

• Serology patterns over age and through time will help demystify the role of children in transmission (linked to core question of school closures) and establish risks to older populations via transmission (models + information on contact patterns will be required).

• Larger sampling (including internationally!) should clarify transmission and burden effects (from socio-economic, to co-morbidities, to genetics).

• Serology in tandem with the step down of Non-Pharmaceutical Interventions could establish the costs and benefits of different interventions.

A

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age

of c

onta

ct

0

2

4

6

8

10+

Regular day

Wuh

an

A

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

0

2

4

6

8

10+

OutbreakB

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

−20246810+

DifferenceC

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

Age

of c

onta

ct

0

2

4

6

8

10+

Shan

ghai

D

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

0

2

4

6

8

10+

E

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

−20246810+

F

All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint (which was not peer-reviewed) is.https://doi.org/10.1101/2020.03.19.20039107doi: medRxiv preprint

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age

of c

onta

ct

0

2

4

6

8

10+

Regular day

Wuh

an

A

05

1015202530354045505560

65+0 5 10 15 20 25 30 35 40 45 50 55 60 65+

0

2

4

6

8

10+

OutbreakB

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

−20246810+

DifferenceC

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

Age

of c

onta

ct

0

2

4

6

8

10+

Shan

ghai

D

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

0

2

4

6

8

10+

E

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

−20246810+

F

All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint (which was not peer-reviewed) is.https://doi.org/10.1101/2020.03.19.20039107doi: medRxiv preprint

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age

of c

onta

ct

0

2

4

6

8

10+

Regular day

Wuh

an

A

05

1015202530354045505560

65+0 5 10 15 20 25 30 35 40 45 50 55 60 65+

0

2

4

6

8

10+

OutbreakB

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

−20246810+

DifferenceC

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

Age

of c

onta

ct

0

2

4

6

8

10+

Shan

ghai

D

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

0

2

4

6

8

10+

E

05

1015202530354045505560

65+

0 5 10 15 20 25 30 35 40 45 50 55 60 65+

Age of participant

−20246810+

F

All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint (which was not peer-reviewed) is.https://doi.org/10.1101/2020.03.19.20039107doi: medRxiv preprint

Diary studies in Wuhan

5. Serology: the use case

• Proportion seronegative at one time-point tells us ‘how many people are left to infect’: where are we on the epidemic trajectory?

• Longitudinal samples that track individuals’ status through time will help establish: what does seropositivity ‘mean’ for immunity?

• Serology patterns over age and through time will help demystify the role of children in transmission (linked to core question of school closures) and establish risks to older populations via transmission (models + information on contact patterns will be required).

• Larger sampling (including internationally!) should clarify transmission and burden effects (from socio-economic, to co-morbidities, to genetics).

• Serology in tandem with the step down of Non-Pharmaceutical Interventions could establish the costs and benefits of different interventions.

5. Serology: the use case

Thank you!

C. Jessica E. Metcalfcmetcalf@princeton.edu

Illustrating model calibration:https://labmetcalf.shinyapps.io/serol1/

Literature overview: https://larremorelab.github.io/covid19testgroup

Combining models and serological data: https://dash.harvard.edu/handle/1/42659939

Resources