|
|
| |||||||||||||||||||||||||||||||||||||||||
Background Maintenance of long-term antibody responses is critical for protective immunity against many pathogens. However, the duration of humoral immunity and the role played by memory B cells remain poorly defined.
Methods We performed a longitudinal analysis of antibody titers specific for viral antigens (vaccinia, measles, mumps, rubella, varicella–zoster virus, and Epstein–Barr virus) and nonreplicating antigens (tetanus and diphtheria) in 45 subjects for a period of up to 26 years. In addition, we measured antigen-specific memory B cells by means of limiting-dilution analysis, and we compared memory B-cell frequencies to their corresponding serum antibody levels.
Results Antiviral antibody responses were remarkably stable, with half-lives ranging from an estimated 50 years for varicella–zoster virus to more than 200 years for other viruses such as measles and mumps. Antibody responses against tetanus and diphtheria antigens waned more quickly, with estimated half-lives of 11 years and 19 years, respectively. B-cell memory was long-lived, but there was no significant correlation between peripheral memory B-cell numbers and antibody levels for five of the eight antigens tested.
Conclusions These studies provide quantitative analysis of serologic memory for multiple antigens in subjects followed longitudinally over the course of more than one decade. In cases in which multiple exposures or repeated vaccinations were common, memory B-cell numbers did not correlate with antibody titers. This finding suggests that peripheral memory B cells and antibody-secreting plasma cells may represent independently regulated cell populations and may play different roles in the maintenance of protective immunity.
Methods
Subjects
We recruited human subjects from the Oregon National Primate Research Center through campuswide e-mail, inviting them to provide serum samples for banking at enrollment and to provide additional serum samples from scheduled blood draws. Fifty-one people responded, providing written informed consent and completing an extensive medical-history questionnaire before providing samples for study. No inclusion criteria were used other than a requirement that the subject have at least three serum samples banked for 3 years or more before the study began. On the basis of this criterion, 6 subjects were excluded, and 45 subjects provided a total of 630 serum samples for analysis.
Samples were drawn annually as part of a centerwide, comprehensive program to permit serologic testing of people working in close proximity with nonhuman primates. In the event of an exposure to an animal (e.g., a scratch or bite), an additional serum sample was drawn. Blood samples were also drawn at weekly intervals after smallpox vaccination in a subgroup of eight subjects in order to monitor the acute phase of the immune response. The study was approved by the institutional review board of Oregon Health & Science University.
ELISA and Memory B-Cell Measurement
Measurement of antibodies by means of enzyme-linked immunosorbent assay (ELISA) was performed as previously described.11,12 Antigens included vaccinia (WR strain prepared in our laboratory), measles (Edmonston strain, Biodesign), mumps (Enders strain, Biodesign), rubella (HPV77 strain, Viral Antigens), varicella–zoster virus (Rod strain, Biodesign), Epstein–Barr virus gp125 (Viral Antigens), tetanus toxin (C-fragment), and diphtheria toxin (EMD Biosciences). International serum standards were obtained from the National Institute for Biological Standards and Controls in Hertfordshire, England. An optimized limiting-dilution assay was used to measure memory B cells.13
Statistical Analysis
Before the final analysis, antibody data were censored by excluding unvaccinated subjects, seronegative subjects, and subjects with test results that were equivocal (<200 ELISA units). A total of one to nine subjects were excluded, depending on the antigen. We removed data points obtained during or shortly after an acute immune response before determining antibody half-life estimates. For example, if more than a doubling of the titer was observed between contiguous time points (hereafter referred to as an antibody spike), then that time point and those from the next 3 years were removed from the analysis so that the rapid initial decrease in antibodies typically observed shortly after antigenic exposure14,15,16,17 would not influence the estimated long-term decay rate. A subject had to have at least three data points for a period of 3 years or more in order to be included in the final analysis. Tetanus and diphtheria vaccines are frequently administered in combination,18 so if an antibody spike was observed between contiguous data points for one antigen, a reexposure was assumed for the reciprocal antigen and removed from the analysis. Since tetanus–diphtheria toxoid revaccination is recommended every 5 to 10 years, contiguous data points separated by more than 5 years were not used for the analysis of these antigens.
To address possible changes before and after acute immune responses, the analysis was performed first for only the longest period of observation and then for all the data, with each period treated as an independent longitudinal series. The results of these two analyses did not differ. For each antigen described in this study, we present the results of the analysis in which all data were used.
Rates of antibody decay were estimated with the use of a longitudinal mixed-effects model (PROC MIXED; SAS version 9.1, SAS Institute). The model had a fixed effect for age and gaussian-distributed random intercepts and slopes (see the Supplementary Appendix, available with the full text of this article at www.nejm.org). Random components were assumed to be uncorrelated. Measures were logarithmically transformed (on a natural log scale) before analysis, resulting in a single-exponent decay model of the untransformed data. Visual inspection indicated that the linear model fitted the data for the subjects well. We assessed the influence of data for any one subject on the analysis with the use of cross-validation. The results were not influenced by any one person's series.
Half-life estimates were obtained by transforming the decay rate and the boundaries of the 95% confidence interval obtained from the fixed-effects slope component of the mixed-effects model. If a specific antibody level does not significantly decay or shows even a small insignificant increase over time (e.g., <1%), then the calculated half-life becomes infinity. To confirm the results from our mixed-effects model, we also fitted least-squares linear regression models to each series. The results from the two approaches were nearly identical (see the Supplementary Appendix). For each antigen, we determined whether the population slope was equal to zero in the mixed-effects model. The P values presented are two-sided, and a P value of less than 0.05 was considered to indicate statistical significance. The comparisons of memory B-cell frequencies and associated serum antibody titers were modeled through least-squares linear regression after log10 transformation.
Results
Serum Samples and Characteristics of the Subjects
A total of 630 serum samples from 45 subjects were obtained for this study or were obtained from the Oregon National Primate Research Center serum bank (Table 1). The majority of these serum samples were obtained from scheduled collections, with 50 samples (7.9%) obtained after an unscheduled event (e.g., exposure to an animal). Each subject provided, on average, 14.0 serum samples (median, 11) during an average period of 15.2 years (median, 15.6). The majority of subjects had received smallpox vaccination during childhood and had recovered from viral infections, including measles, mumps, rubella, Epstein–Barr virus, and varicella–zoster infections. The subjects in this cohort had common coexisting conditions but no specific immune deficiencies (Table 1 of the Supplementary Appendix). The duration of serum antibody production was determined with the use of a mixed-effects model of longitudinal analysis. Overall, antigen-specific antibody responses were long-lived, and we found no significant differences in antibody-maintenance patterns according to sex (Table 2).
|
|
One potential mechanism for maintaining long-term immunity to a nonpersisting pathogen is through intermittent reexposure. To examine this possibility, we evaluated antibody responses after smallpox vaccination (Figure 1A). A total of 39 of 43 subjects born before 1972 (91%) were seropositive for vaccinia (Fig. 1A in the Supplementary Appendix), which is consistent with the findings of a previous study.11 Before 2003, when eight subjects received booster vaccination, there were only two instances of an antibody spike that was indicative of vaccination or exposure (0.3 event per 100 person-years). This finding is consistent with the discontinuation of routine vaccination in 1972 and the absence of endemic orthopoxviruses in North America that are known to infect humans. Excluding serum samples obtained after smallpox vaccination in 2003, a putative protective level of antiviral antibody11,21 was present in 28 of the 45 subjects (62%), and the level of vaccinia-specific antibodies decreased slowly, with an estimated half-life of 92 years (95% confidence interval [CI], 46 to infinity; P=0.049).
|
Measles is rare in the United States, but it remains a serious threat through importation.22 An analysis of measles-specific antibodies revealed that five subjects had spikes in serum antibody levels (Fig. 1B in the Supplementary Appendix). Two subjects received documented measles–mumps–rubella (MMR) vaccinations; one seroconverted, whereas the other seropositive subject had antibody titers that did not change after vaccination. Four subjects unknowingly contracted a cross-reactive but uncharacterized paramyxovirus infection from exposure to diseased nonhuman primates during a 1999 outbreak. On the basis of the antibody titers from the last available blood sample drawn before a recent MMR vaccination or serologic boosting during the 1999 outbreak of primate paramyxovirus, 43 of the 45 subjects (96%) were seropositive, with a putative protective level of at least 0.2 IU of antimeasles antibodies per milliliter23 in 41 subjects (91%) (Fig. 1B in the Supplementary Appendix). The decrease in measles-specific antibodies (Figure 1B) was not significant (P=0.94) and is likely to be maintained for life (estimated half-life, 3014 years; 95% CI, 104 to infinity).
Mumps
Although typically less severe than measles, mumps is another childhood disease with the potential for serious complications.18 In this cohort, 41 of 45 subjects (91%) were seropositive for mumps (Fig. 1C in the Supplementary Appendix). Two subjects received MMR vaccinations during the period of observation, and two other subjects had spikes in antibody levels, one in 1998 and the other in 2004. These spikes may have been the consequence of exposure to naturally occurring mumps rather than MMR vaccination. This incidence rate of 0.3 event per 100 person-years is consistent with a steady decrease in prevalence18 since routine vaccination began in 1977. Antibody responses to mumps (Figure 1C), like those to measles, were long-lived (estimated half-life, 542 years; 95% CI, 90 to infinity) and showed no significant decrease (P=0.69).
Rubella
Rubella was a leading cause of birth defects in the United States before immunization programs were implemented in the 1970s.18 A total of 40 of the 45 subjects in our cohort (89%) were seropositive for rubella (Fig. 1D in the Supplementary Appendix). We identified two subjects with documented MMR vaccinations and one subject with a spike in preexisting rubella-specific titers in 2003. Because of an 8-year gap between contiguous serum samples, it is unclear whether the change in this latter subject was due to a natural case of rubella or variability in serum titers. In any case, a spike in rubella-specific antibodies is uncommon, with a low overall incidence rate of 0.15 event per 100 person-years. A putative protective level of at least 10 IU of antirubella antibodies per milliliter24 was reached in 39 of the 45 subjects (87%). Rubella-specific immunity (Figure 1D) was maintained, with an estimated half-life of 114 years (95% CI, 48 to infinity) and no significant rate of decrease (P=0.15).
After reviewing the medical histories of the subjects, we identified six subjects in whom only vaccine-induced immunity against measles had developed, four subjects in whom vaccine-induced immunity against mumps had developed, and seven subjects in whom vaccine-induced immunity against rubella had developed. Exclusion of these subjects from the longitudinal analysis shown in Figure 1 had no substantial effect on the calculated duration of serum antibody responses (Table 2 of the Supplementary Appendix). Moreover, although reexposure to the same or a serologically cross-reactive virus may boost antibody titers, these data with regard to vaccinia, measles, mumps, and rubella indicate that repetitive environmental exposures and infections are not absolutely required for maintaining long-term antiviral antibody responses.
Epstein–Barr Virus and Varicella–Zoster Virus
In contrast to acute viral infections, chronic and latent viral infections may either persist or be reactivated from latency, thereby "boosting" immune responses in the infected person. Antibody titers were determined for two latent herpesviruses, Epstein–Barr virus and varicella–zoster virus. A total of 37 of the 45 subjects (82%) were seropositive for Epstein–Barr virus, with seroconversion in 1 subject occurring during the observation period (Fig. 1E in the Supplementary Appendix). Four seropositive subjects had antibody titers that spiked during observation, indicating that reactivation or reexposure events had occurred, but at a relatively low rate (0.6 event per 100 person-years). Humoral immunity against Epstein–Barr virus (Figure 1E) showed no significant decrease (P=0.99) and is likely to be maintained for life (estimated half-life, 11,552 years; 95% CI, 63 to infinity).
Unlike antibody responses to Epstein–Barr virus, antibody responses to varicella–zoster virus showed frequent fluctuations (Fig. 1F in the Supplementary Appendix). All 45 subjects were seropositive for varicella–zoster virus, and 10 of the 45 subjects (22%) had antibody spikes (1.6 events per 100 person-years). Two subjects described an episode of shingles at or near the time of the observed spike in antibody responses to varicella–zoster virus, one subject may have been exposed to recently vaccinated children, six subjects do not recall having shingles or any known exposure to patients with varicella–zoster virus, and information was not available for one subject. Immunity (Figure 1F) decreased slowly, with an estimated half-life of 50 years (95% CI, 30 to 153; P=0.005). Thus, although the infection is latent with evidence of the most frequent reexposure and reactivation events, varicella–zoster virus induced the most short-lived antibody response of the viruses we examined.
Tetanus and Diphtheria
Tetanus and diphtheria vaccination has been recommended since the 1940s, primarily with the combined tetanus–diphtheria toxoid vaccine for adults. This recommendation has resulted in a sharp decrease in the incidence of both diseases18 and prolonged maintenance of antibody titers (Figure 1G and 1H). A titer of more than 0.01 IU of antitetanus antibodies per milliliter is considered to be protective.25 ELISA titers of more than 0.16 IU per milliliter correlate well with neutralizing activity; a titer of 0.16 IU per milliliter is the lowest level reliably detected by means of ELISA26 and is similar to our detection limit of 0.15 IU per milliliter (200 ELISA units). Protective antitetanus responses (Fig. 1G in the Supplementary Appendix) were clearly identified in 42 of the 45 subjects (93%), but this response rate may not reflect the true number of protected persons, since 0.01 IU is below the limit of detection by ELISA. Frequent tetanus boosters resulted in 31 instances of an antibody spike in 27 subjects (4.9 events per 100 person-years) (Fig. 1G in the Supplementary Appendix). Tetanus-specific antibodies decreased rapidly, with an estimated half-life of 11 years (95% CI, 10 to 14; P<0.001), which is similar to the decrease shown in a model reported more than 40 years ago.27
To determine whether the rapid antibody decay observed with tetanus held true for other protein antigens, we measured immunity against diphtheria (Figure 1H). Antidiphtheria antibody titers of more than 0.01 IU per milliliter are considered to be protective,28 and in 40 of the 45 subjects (89%) diphtheria-specific antibodies remained above our detection limit (200 ELISA units, or 0.04 IU per milliliter) (Fig. 1H in the Supplementary Appendix). It is possible that the remaining five subjects also had sustained protective antidiphtheria immunity but that it could not be definitively measured in these studies. Diphtheria-specific antibody spikes were observed, typically in parallel to responses against tetanus, as expected because of the combined tetanus and diphtheria formulation recommended for adult vaccination. An analysis of diphtheria-specific antibodies indicated an estimated half-life of 19 years (95% CI, 14 to 33; P<0.001). These results suggest that antibody maintenance is greatly influenced by the nature of the antigen, with these proteins eliciting quantitatively shorter antibody responses than those observed after viral infection.
Antibody Maintenance in Individual Subjects
One model of antibody maintenance predicts that long-term responses are maintained by non–antigen-specific stimulation, also referred to as "bystander activation" of memory B cells during antigenically unrelated infections.3,8,29 To determine the potential effects of heterologous infection and vaccination, we measured humoral responses to eight antigens in four subjects followed longitudinally for more than 25 years and at weekly intervals after smallpox vaccination (Figure 2). Common immunologic events, including tetanus and diphtheria booster immunization (in Subjects 1 through 4), Epstein–Barr virus seroconversion (in Subject 2), and varicella–zoster virus reexposure or reactivation (in Subject 4), occurred during the period of observation but showed little effect on other antigen-specific antibody responses. After a defined infection with vaccinia (i.e., booster smallpox vaccination), there was little or no alteration in antibody responses to seven other antigens (<6% average change) (Figure 2), despite vaccinia-specific antibody responses that increased by approximately 4000% at the peak of the response. These results are consistent with those in six subjects who received booster smallpox vaccination and two subjects who received primary smallpox vaccination and later received either live yellow-fever vaccination or MMR vaccination (data not shown). Together, these findings suggest that nonspecific bystander activation is an unlikely mechanism in the maintenance of long-term antibody responses.
|
B-Cell Memory and Association with Antibody Levels
Plasma cells either maintain antibody levels independently30,31,32 or may require replenishment by the proliferation and differentiation of memory B cells. A requirement for all memory B-cell–dependent theories of antibody maintenance is that a correlation must exist between memory B-cell levels and antibody levels.3,8,29 With the use of a limiting-dilution analysis,13 we found that memory B cells in the circulation were remarkably long-lived (Figure 3A). Previous studies have shown that within 1 month after vaccination, memory B-cell numbers in the circulation are representative of the memory B-cell frequencies observed in other lymphoid compartments such as the spleen.33
|
|
We investigated the longitudinal maintenance of antibody responses to eight antigens over a period of up to 26 years. Antibody responses after live viral infections had half-lives of 50 years or more, with many showing no measurable decrease. In contrast, responses to nonreplicating protein antigens (tetanus and diphtheria) decreased at a relatively rapid pace (estimated half-lives, 11 to 19 years), suggesting that within a particular person, antigen-specific mechanisms play a substantial role in determining the duration of humoral immunity.
On average, subjects were 52 years of age at the conclusion of the study, and most had contracted natural measles, mumps, or rubella infections during childhood. It is unknown whether vaccine-induced immunity is as long-lived as that induced by natural infection. Although the titers of vaccine-induced antibody responses appear to be lower than those reached after natural infection (data not shown), a longitudinal analysis of immunity against measles, mumps, and rubella suggests that serologic memory may be similar to natural infection (Supplementary Appendix). However, this result is based on a very small sample (two to five subjects), and further studies on antibody maintenance will be required before statistically meaningful comparisons can be made. This information is particularly important in contemporary populations in which asymptomatic reboosting of immunity by circulating natural or wild-type viruses appears to be relatively rare.
Two principal mechanisms for maintaining long-term humoral immunity have been proposed: memory B-cell–dependent antibody production by short-lived plasma cells and memory B-cell–independent antibody production by long-lived plasma cells.3,4,5,6,7,8,9,10 In memory B-cell–dependent models, memory B cells divide and differentiate into plasma cells after stimulation by persistent antigen, reinfection, cross-reactive antigens, or non–antigen-specific polyclonal activation. If polyclonal memory B-cell activation is the key to antibody maintenance,29 then the duration of antibody production should be the same, regardless of antigen specificity.
Moreover, hypotheses of memory B-cell dependence require that after each infection, a correlation between memory B cells — or at least peripheral memory B cells29 — and antibodies be maintained. This correlation has been noted in some instances35,36 but not in others.37,38 We found an association between memory B-cell levels and antibody levels for measles, mumps, and rubella but not for vaccinia, varicella–zoster virus, Epstein–Barr virus, tetanus, or diphtheria (Table 3). The associations with these first three viruses may reflect an epiphenomenon in which serum antibody levels and memory B cells are equally stable but independently maintained, without representing a direct cause-and-effect relationship.
If plasma-cell maintenance is memory B-cell–dependent, then booster vaccination or reinfection may reveal whether a cause-and-effect relationship is involved, because the association will continue to be maintained. However, if memory B cells and plasma cells are independently regulated, then multiple reexposures to antigens may cause divergence between memory B-cell levels and antibody levels. In support of this second hypothesis, antigens with the highest rates of boosting through vaccination or latent viral infection coincidentally showed the weakest association between memory B-cell titers and antibody titers (Table 3). Similarly, another study showed that multiple tetanus immunizations lead to a sustained increase in memory B cells without a concomitant increase in long-term antibody titers, indicating that memory B cells and antibody production are independently regulated.38
Long-lived plasma cells are another mechanism for maintaining serum antibodies. The existence of long-lived plasma cells was first shown in mice,30,31,32 and preliminary studies in human subjects depleted of CD20+ B cells provide further support for the long-lived plasma-cell hypothesis.39,40
Our cohort had seroprevalence rates of 82 to 100%, depending on the antigen being tested. All subjects had a response to at least five of the eight antigens tested; thus, no subjects had an overt inability to mount long-lasting antibody responses. Participants in this study were in good health overall, although certain persons had coexisting conditions such as high blood pressure or asthma (Table 1 in the Supplementary Appendix). Unrecognized confounders, including immune deficiencies and ablative treatments such as chemotherapy, may affect humoral immunity, resulting in more rapidly decaying antibody responses than those that we observed. Likewise, our findings are based on T-cell–dependent antibody responses, and the mechanisms involved in antibody persistence may differ for humoral immunity against T-cell–independent antigens, which is often short-lived. Greater insight into the factors that determine the duration of specific antibody responses will be important for future vaccine design, as well as for determining the timing of booster vaccinations required to sustain protective levels of immunity.
Supported by grants (PO1 AG023664, RR000163, RR000334, and RR024140) from the Public Health Service.
No potential conflict of interest relevant to this article was reported.
We thank the study subjects for their time and their participation in this study, J. Hall and D. Hess for obtaining serum samples, S. Carter for aid with organizing serum samples, and M. Lewis and E. Hammarlund for blood collection and processing.
Source Information
From the Vaccine and Gene Therapy Institute, Oregon Health & Science University, Beaverton (I.J.A., M.K.S.); and the Department of Public Health and Preventative Medicine, Division of Biostatistics, Oregon Health & Science University, Portland (N.E.C.).
Address reprint requests to Dr. Slifka at the Vaccine and Gene Therapy Institute, Oregon Health & Science University, 505 NW 185th Ave., Beaverton, OR 97006, or at slifkam{at}ohsu.edu.
References
| |||||||||||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
