Long term intrinsic cycling in human life course antibody responses to influenza A(H3N2): an observational and modeling study

Article activity feed

1. 

   eLife

   Dec 6, 2022

   Author Response

   eLife Assessment:

   This manuscript follows the still unanswered concept of 'original antigenic sin' and shows the existence of a 24-year periodicity of the immune response against influenza H3N2. The valuable work suggests a long-term periodicity of individual antibody response to influenza A (H3N2) within a city. But, to substantiate their argument, the authors would need to provide additional supporting data.

   Thank you for your comments. We have performed additional analyses and included those results in the revision to support our findings.

   Specifically, we included a sensitivity analyses that predicting phases by fitting models with 35- and 6-years periodicity, which were found to provide poorer predictions than the 24-year periodicity used in our main results (Figure 4 – figure supplementary 1).

   We also generated …

   Author Response

   eLife Assessment:

   This manuscript follows the still unanswered concept of 'original antigenic sin' and shows the existence of a 24-year periodicity of the immune response against influenza H3N2. The valuable work suggests a long-term periodicity of individual antibody response to influenza A (H3N2) within a city. But, to substantiate their argument, the authors would need to provide additional supporting data.

   Thank you for your comments. We have performed additional analyses and included those results in the revision to support our findings.

   Specifically, we included a sensitivity analyses that predicting phases by fitting models with 35- and 6-years periodicity, which were found to provide poorer predictions than the 24-year periodicity used in our main results (Figure 4 – figure supplementary 1).

   We also generated a antigenic map with the locations of our tested strains shown in the map. We also compared the paired antigenic distance of A(H3N2) strains (including our tested strains). These results (Figure 1 – figure supplementary 3) suggested that the tested strains that we used spanned the circulation of A(H3N2) since its emergence and well covered the antigenic space of the virus.

   Reviewer #1 (Public Review):

   The authors suggest that there is a long-term periodicity of individual antibody response to influenza A (H3N2). The interesting periodicity may be surely appeared. Though the authors assume that the periodicity is driven by pre-existing antibody responses, the authors could provide more supportive data and discuss some possibilities.

   Thank you for your comments and please find our point-to-point responses below.

   1. The authors can investigate whether the periodicity reflects an epidemic/invasion record of A(H2N3) within Guangzhou or the surrounding city, e.g., the numbers of flu-infected people yearly can be referred to.

   Thank you for your comments. We aimed to investigate the periodicity in individual level antibody responses, so we made several efforts to minimize the impacts of population level A(H3N2) activity in our analyses. In particular, we have removed the average activity at population level (i.e., strain-specific intercepts), to minimize the impact of higher circulation of a certain stain on the periodicity.

   In our simulations, we tested models that only incorporated population level activity but not including cross-reactions (Figure 3B, I), which did not recover the observed periodicity. In the models that including both population level activity and cross-reactions, we found that less predictable population level activities (i.e., less regular annual epidemics) would increase the variations in individual-level long-term periodicity (Figure 3G-H). We also found that measured periodicities did not vary substantially when comparing those measured at baseline compared to those measured at follow up (~3-4 years later). These results suggested that the local epidemics may only have limited impacts on the observed periodicity in individual’s antibody responses, while the cross-reactions between previous exposed and currently circulating strains may be the main drivers.

   To address this comment, we added a paragraph in discussion (lines 336-342):

   “In this study, we did not explore the interactions between individual level antibody responses with population level A(H3N2) activity (e.g., epidemic sizes). We minimized the impacts from population level by performing the Fourier analysis with individual departures from population average and validating the results with data from the Vietnam cohort. Simulation results further suggested that the population level virus activity alone was not able to recover the observed periodicity, though epidemics with less regularity seemed to increase the variability in individual-level periodicity in the presence of broad cross-reactions (Figure 3G-H).”

   1. The authors can consider whether the participants are recently/previously vaccinated and/or infected with flu. The remaining antibodies may reflect a long memory but may show a recent activation.

   Thank you for your comments. We agree with the reviewer that the observed seroconversion of the circulating strains may reflect responses recent re-exposures. Given the low influenza vaccine coverage in our cohort (1.3%, 10 out of 777) and in China in general (<5% [3, 4]), we believe that our observed periodicity and seroconversion patterns were unlikely to be caused by to recent influenza vaccinations.

   We think that the pervasive exposure to A(H3N2) could be a driver to the observed seroconversions to circulating strains between our baseline and follow-up were likely due to the pervasive exposures (or reinfections for those who developed into infections). Using the same data set, we previously reported 98% and 74% of participants experienced 2- and 4-fold rise to any of the 21 tested A(H3N2) strains [5].

   As the reviewer and previous studies suggested, the antibody responses could reflect long term memories that were activated after recent exposures [1, 6]. We generated our hypothesis based on this features, and to characterize the periodicity that may arose from the interactions between long term memories and newly generated antibodies.

   We incorporate the re-infection mechanism in our simulations, with and without subsequent cross-reactions with previously exposed distant strains (Figure 3I). Results indicate that reinfection alone cannot recover the observed long-term periodicity (Figure 3A), while reinfection plus the resulting cross-reactions can recover such long-term periodicity (Figure 3D). Therefore, we believe that the repeated exposures or re-infections would not affect our reported periodicity, while they may be drivers of continuous formulation of the life-course antibody profiles and the observed periodicity. Of particular note is the consistency of measured periodic behaviour at baseline and follow up (~3-4 years later).

   To address this comment, we reported the vaccination status of our participants when introducing the data (lines 127-129) and in the discussions (lines 280-282 and 313-315):

   “Only 0.6% (n = 5) of participants self-reported influenza vaccinations between the two visits, therefore, the observed changes in HI titers between the two visits were likely due to natural exposures.”

   “Due to the low influenza coverage in our participants and in China in general, the observed seroconversions likely reflected antibody responses after natural exposures during the study period.”

   “Particularly, our simulation results suggested that model including repeated exposures or population level A(H3N2) activity alone did not recover the long-term periodicity (Figure 3).”

   1. The strains inducing high HI titers may have similar mutations and may be reactive to the same antibodies. What are the mutation frequencies among 21 A(H3N2) strains?

   Thank you for your comments. We selected the 21 tested strains to cover the span of the circulation of A(H3N2) strains since 1968 and antigenic diversity. We prioritized with the strains that were included in the vaccine formulation and tested to create the antigenic map by Fonville et al. [1].

   We reproduced the antigenic map (up to strains isolated in 2010) by Fonville et al. [1] and compared the antigenic locations of our tested A(H3N2) strains (Figure 1—figure supplement 3). The 21 strains (or their belonging antigenic clusters if the strains were not used for the map) largely tracked the antigenic evolution of A(H3N2) since its emergence in 1968, with a reportedly mutation rate of 0.778-unit changes in antigenic space per year [1, 2].

   We further calculated the paired antigenic distance of strains tested in the antigenic map, which was highly correlated with the time intervals between the isolation of the two strains. The figure also suggested our tested strains cover the time spans and antigenic distances that were shown in the original antigenic map. In addition, our observed periodicity was identified in individual time series of residuals, which has removed the shared virus responses or assay measurements (Figure 1). Therefore, we believe that the impact of specific mutations may have limited impacts on our findings.

   To address this comment, we included the reproduced antigenic map showing the locations of the tested strains and their pair-wise antigenic distance in Figure 1—figure supplement 3 and referenced in the main text (line 127).

   Reviewer #2 (Public Review):

   This is a well-thought-out, clearly exposed article. It builds upon the platform of 'original antigenic sin' (OAS), a notion first developed from studying individuals infected with influenza. According to OAS, the initial infection will set the dominant immune response targets (antigens) that immune cells will recognize, such that infection with a related strain will cause a strong response focused mainly against the initially infecting strain, that then goes on to protect against the new-infecting strain. This study builds off this idea, showing that as strains become increasingly antigenically distant as inferred by the time between strain appearance, the cross-protection can drop to a point where it needs to be invigorated with a potentially new response. The potential biological mechanisms behind this aren't discussed, but a model is built that conveys the potential for 'relative risk' of an individual over the course of the life, based essentially on when one was born.

   Thank you for your comments. We expanded our introduction hoping to include more biological mechanisms, especially those related with original antigenic sin.

   “Antibodies mounted against a specific influenza virus decay (in either absolute magnitude or antigenic relevance) after exposure until re-exposure or infection to an antigenically similar virus occurs, whereupon back-boosting of antibodies acquired from previous infections (e.g., activation of memory B cells) can occur, as well as updating antigen specific antibodies to the newly encountered infection (e.g., activation of naïve B cells.” (lines 80-84)

   “Original antigenic sin (OAS) is a widely accepted concept describing the hierarchical and persistent memory of antibodies from the primary exposure to a pathogen in childhood. Recent studies suggested that non-neutralizing antibodies acquired from previous exposures can be boosted and may blunt the immune responses to new influenza infections.” (lines 92-97)

   The basic premise was to measure from serum influenza haemagglutinin-inhibition (HI) titers of 21 strains of influenza A (H3N2) - related strains causing disease at various times over a period of some 40 years- from a diverse set of ≈800 participants of various ages, at two time points, spaced 2 yr apart. The authors then calculated the HI titer for the 21 strains for each individual. From this, each participant's age, their age at the time of a strain's development, and when a strain emerged were used to assess whether there was periodicity to immune responses by performing a splined Fourier transform for each individual and then examining the composite pattern across time for HI titers. The authors propose that on average there is a 24-year periodicity to immune responses to influenza strains, such that after the initial infection, cross-reactivity reduces to the point where it may be less meaningful for protection over around 24-year, and suggests activation of a 'new' immune response might be required to control the more distant strain involved in the response at that time. The periodicity was longer than would be predicted if age were not a factor involved in the HI titer patterns across time. Further, variability in the periodicity was shown to involve broad cross-reactivity between strains and narrow cross-reactivity in more highly-related (closer in time) strains, individual HI titer, and periodic population fluctuations. In the literature, viral strains are estimated to mutate to the point of losing 50% cross-reactivity with a T1/2 of approximately 2.5 yr, which would make the inferred lifespan plausible but perhaps surprisingly long, implying there are immune feedback parameters that influence periodicity. The authors also use an independent cohort of approximately 150 individuals from a separate, published, study to validate some findings revealed in the primary data set.

   Thank you for your comments and sorry for the confusion. We agree with the reviewer that the onward protection from the cross-protection should be shorter than 24-year periodicity that was identified in the retrospective antibody responses. We hope to clarify that we identified long-term periodicity by retrospectively investigating the individual antibody profiles, which were results of multiple previous exposures and immunity and cross-reactions that arose from these previous exposures. Therefore, the long-term periodicity is a retrospective characterization, and should not be directly interpretated as the duration of onward protection.

   As shown in Figure 4A, the 24-year periodicity consists of phases when individuals’ titers are higher (phase I & II) and lower (phase III & IV) than the population average. As such, the duration of onward protection may be shorter than the entire periodicity. Assuming the protection decreasing with lower titer levels, the onward protection is expected to decrease in phase II and take 1-6 years to drop from the furthest to population average. This is consistent with findings that homotypic cross-protection against PCR-confirmed infections up to about five seasons (lines 291-293), but whether such protection is driven by the declining of cross-reactions still need further investigations.

   To address this comment, we rephrased our discussion and make the interpretation less confusing. (lines 285-287):

   “Of note, the long-term periodicity is a retrospective characterization of individual antibody profiles that arose from multiple exposures and cross-protection, which should not be directly interpreted as the duration of onward protection conferred by the existing antibodies.”

   Strengths: Overall, the study is well executed and the patterns that are visually apparent in Figure 1A (the 'raw' data) are built on to inform a model of the potential breadth of cross-reactivity in a given individual at any given time after birth, integrated with the influenza strains to which they are most likely to have been first exposed. It is a complex thing to make sense of data involving many individuals who could be infected or vaccinated at any and variable points in time over the course of their life, but the authors derive a model that probabilistically accounts for possible infection events, so controls for this nicely, or at least to a degree that is practicable.

   Thank you for your supportive comments. We hope to clarify that we identified the long-term periodicity using the residuals of individual HI titers after extracting the population activity that is visually noticeable in Figure 1A. By doing this, we hope to minimize the impacts of population level A(H3N2) activity and laboratory measurements on individual antibody responses (Figure 1C; detailed methods in lines 396-412).

   Questions related to the main limitation: The level of math in this paper makes it hard for a basic biologist to critique the approach, but the argued points are intriguing. Foremost, in the final part of the paper the authors move from building a model to testing its potential to predict HI titers in the final quarter strains of the study period, placing individuals into one of four phases: I) early increasing to high titer response, II) waning response phase where they are returning back to the average population-level response against a strain, III) sub-par response against a strain and then reinitiation of HI titers in phase IV. Pleasingly this shows a good correlation between individuals' ages and their predicted phase. However, while the fit predicts phase well in Fig 4C and 4D, it looks to perform less adequately in Fig 4B.

   1. Why is this?

   Thank you for your comments and sorry for the confusion. In Figure 4B, we aimed to characterize and predict the position instead of the amplitude in the individual time series of residuals. Therefore, we fitted the model using only harmonic terms (i.e., sine and cosine functions; Equation 12 on page 26) [7], while we believe there may be other factors that could affect the observations but were not included in the model. The perditions from the model inform the position and velocity of harmonic oscillators rather than the amplitude or extent of the wave, therefore, the predictions did not exactly fit the observations.

   To address this comment, we expand the corresponding methods hoping to make it clear (lines 661-663):

   “Of note, we fitted the model aiming to estimate the position of the harmonic oscillators and did not consider for other non- harmonic factors, therefore the model may not fully capture the variations of the data.”

   1. Another point for consideration is that the time between samplings (2010-2012) is comparatively short, given a 24-yr predicted periodicity. What would happen to the predictions if the periodicity were 35-yr or 6-yr? Would the model fail to call individuals accurately in these cases?

   Thank you for your comments. We repeated our predictions in Figure 4F-G by assuming a 35-year and 6-year periodicity respectively as suggested. Results suggested that model predictions with either 35-year or 6-year did not outcompete the model predictions assuming a 24 years old (Figure 4—figure supplement 1). For instance, the observed proportion of seroconversion to circulating strains in each cohort have correlation coefficients of 0.49 (p-value = 0.05), 0.63 (p-value = 0.02) and -0.12 (p-value = 0.69) with the predicted proportion of phase IV when assuming a 35-, 24- and 6-year periodicity, respectively.

   We also hope to clarify that we investigated the prediction potentials of long-term periodicity from two perspectives. Except for using the periodicity to predict the seroconversions between baseline and follow-up, we also predict the phase of each individual in the year of 2012 only using HI titers against strains that were isolated before 2002. Our results suggested our 10-years ahead predictions well correlated with observations (Figure 4C).

   To address this comment, we also included the results of analyses using alternative 35- and 6-year periodicity as Figure 4—figure supplement 1, and reported in the main text (lines 262-264).

   1. Similarly, if the samples were taken further apart, would the model still be effective at predicting phase?

   Thank you for your comments. We hope to clarify that we collected two cross-sectional serum samples, while we identified the long-term periodicity and predicted phase with serums collected from each visit, separately. For instance, in our sensitivity analysis that using serum collected in follow-up (Figure 1—figure supplement 1), we revealed similar long-term periodicity (baseline in Figure 1) with that identified using the baseline serums, despite pervasive exposures during this time period (time separating samples varied from 3-4 years). In addition, the Vietnam data collected sera from six consecutive years. These data showed a similar long-term periodicity (Figure 2—figure supplement 5).

   For the phase prediction, we used residuals of HI titers against 14 historical strains that were isolated between 1968 and 2002, and predicted the phase of strain that was isolated in the year 2012. This prediction was derived purely by depending on the periodic pattern of the time series and without information for strains isolated 10 years prior to 2012. Therefore, the prediction was 10 years ahead and was well correlated with observations from the complete time series, further supporting that there may be an intrinsic cycling in individual antibody responses and that this cycle is fairly stationary and predictable.

   Read the original source
2. 

   eLife

   Dec 6, 2022

   eLife assessment

   This manuscript follows the still unanswered concept of 'original antigenic sin' and shows the existence of a 24-year periodicity of the immune response against influenza H3N2. The valuable work suggests a long-term periodicity of individual antibody response to influenza A (H3N2) within a city. But, to substantiate their argument, the authors would need to to provide additional supporting data.

   Read the original source
3. 

   eLife

   Dec 6, 2022

   Reviewer #1 (Public Review):

   The authors suggest that there is a long-term periodicity of individual antibody response to influenza A (H3N2). The interesting periodicity may be surely appeared. Though the authors assume that the periodicity is driven by pre-existing antibody responses, the authors could provide more supportive data and discuss some possibilities.

   1. The authors can investigate whether the periodicity reflects an epidemic/invasion record of A(H2N3) within Guangzhou or the surrounding city, e.g., the numbers of flu-infected people yearly can be referred to.

   2. The authors can consider whether the participants are recently/previously vaccinated and/or infected with flu. The remaining antibodies may reflect a long memory but may show a recent activation.

   3. The strains inducing high HI titers may have similar mutations and …

   Reviewer #1 (Public Review):

   The authors suggest that there is a long-term periodicity of individual antibody response to influenza A (H3N2). The interesting periodicity may be surely appeared. Though the authors assume that the periodicity is driven by pre-existing antibody responses, the authors could provide more supportive data and discuss some possibilities.

   1. The authors can investigate whether the periodicity reflects an epidemic/invasion record of A(H2N3) within Guangzhou or the surrounding city, e.g., the numbers of flu-infected people yearly can be referred to.

   2. The authors can consider whether the participants are recently/previously vaccinated and/or infected with flu. The remaining antibodies may reflect a long memory but may show a recent activation.

   3. The strains inducing high HI titers may have similar mutations and may be reactive to the same antibodies. What are the mutation frequencies among 21 A(H3N2) strains?

   Read the original source
4. 

   eLife

   Dec 6, 2022

   Reviewer #2 (Public Review):

   This is a well-thought-out, clearly exposed article. It builds upon the platform of 'original antigenic sin' (OAS), a notion first developed from studying individuals infected with influenza. According to OAS, the initial infection will set the dominant immune response targets (antigens) that immune cells will recognize, such that infection with a related strain will cause a strong response focused mainly against the initially infecting strain, that then goes on to protect against the new-infecting strain. This study builds off this idea, showing that as strains become increasingly antigenically distant as inferred by the time between strain appearance, the cross-protection can drop to a point where it needs to be invigorated with a potentially new response. The potential biological mechanisms behind this …

   Reviewer #2 (Public Review):

   This is a well-thought-out, clearly exposed article. It builds upon the platform of 'original antigenic sin' (OAS), a notion first developed from studying individuals infected with influenza. According to OAS, the initial infection will set the dominant immune response targets (antigens) that immune cells will recognize, such that infection with a related strain will cause a strong response focused mainly against the initially infecting strain, that then goes on to protect against the new-infecting strain. This study builds off this idea, showing that as strains become increasingly antigenically distant as inferred by the time between strain appearance, the cross-protection can drop to a point where it needs to be invigorated with a potentially new response. The potential biological mechanisms behind this aren't discussed, but a model is built that conveys the potential for 'relative risk' of an individual over the course of the life, based essentially on when one was born.

   The basic premise was to measure from serum influenza haemagglutinin-inhibition (HI) titers of 21 strains of influenza A (H3N2) - related strains causing disease at various times over a period of some 40 years- from a diverse set of ≈800 participants of various ages, at two time points, spaced 2 yr apart. The authors then calculated the HI titer for the 21 strains for each individual. From this, each participant's age, their age at the time of a strain's development, and when a strain emerged were used to assess whether there was periodicity to immune responses by performing a splined Fourier transform for each individual and then examining the composite pattern across time for HI titers. The authors propose that on average there is a 24-year periodicity to immune responses to influenza strains, such that after the initial infection, cross-reactivity reduces to the point where it may be less meaningful for protection over around 24-year, and suggests activation of a 'new' immune response might be required to control the more distant strain involved in the response at that time. The periodicity was longer than would be predicted if age were not a factor involved in the HI titer patterns across time. Further, variability in the periodicity was shown to involve broad cross-reactivity between strains and narrow cross-reactivity in more highly-related (closer in time) strains, individual HI titer, and periodic population fluctuations. In the literature, viral strains are estimated to mutate to the point of losing 50% cross-reactivity with a T1/2 of approximately 2.5 yr, which would make the inferred lifespan plausible but perhaps surprisingly long, implying there are immune feedback parameters that influence periodicity. The authors also use an independent cohort of approximately 150 individuals from a separate, published, study to validate some findings revealed in the primary data set.

   Strengths: Overall, the study is well executed and the patterns that are visually apparent in Figure 1A (the 'raw' data) are built on to inform a model of the potential breadth of cross-reactivity in a given individual at any given time after birth, integrated with the influenza strains to which they are most likely to have been first exposed. It is a complex thing to make sense of data involving many individuals who could be infected or vaccinated at any and variable points in time over the course of their life, but the authors derive a model that probabilistically accounts for possible infection events, so controls for this nicely, or at least to a degree that is practicable.

   Questions related to the main limitation: The level of math in this paper makes it hard for a basic biologist to critique the approach, but the argued points are intriguing. Foremost, in the final part of the paper the authors move from building a model to testing its potential to predict HI titers in the final quarter strains of the study period, placing individuals into one of four phases: I) early increasing to high titer response, II) waning response phase where they are returning back to the average population-level response against a strain, III) sub-par response against a strain and then reinitiation of HI titers in phase IV. Pleasingly this shows a good correlation between individuals' ages and their predicted phase. However, while the fit predicts phase well in Fig 4C and 4D, it looks to perform less adequately in Fig 4B.

   Q1: Why is this?

   Another point for consideration is that the time between samplings (2010-2012) is comparatively short, given a 24-yr predicted periodicity. Q2: What would happen to the predictions if the periodicity were 35-yr or 6-yr? Would the model fail to call individuals accurately in these cases?

   Q3: Similarly, if the samples were taken further apart, would the model still be effective at predicting phase?

   Read the original source
5. 

   Version published to 10.7554/elife.81457 on eLife

   Dec 2, 2022
6. 

   Version published to 10.1101/2022.06.27.22276898v1 on medRxiv

   Jun 27, 2022