Dr. Stohl has received consulting fees (less than $10,000 per year) from Human Genome Sciences. Dr. Leandro has received honoraria (less than $10,000 per year) from Roche. Dr. Hilbert owns HGSI stock. Dr. Edwards has received consulting fees (less than $10,000 per year), staff support, and a supply of rituximab from Roche.
Circulating levels of B lymphocyte stimulator in patients with rheumatoid arthritis following rituximab treatment: Relationships with B cell depletion, circulating antibodies, and clinical relapse
Article first published online: 28 FEB 2006
Copyright © 2006 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 54, Issue 3, pages 723–732, March 2006
How to Cite
Cambridge, G., Stohl, W., Leandro, M. J., Migone, T.-S., Hilbert, D. M. and Edwards, J. C. W. (2006), Circulating levels of B lymphocyte stimulator in patients with rheumatoid arthritis following rituximab treatment: Relationships with B cell depletion, circulating antibodies, and clinical relapse. Arthritis & Rheumatism, 54: 723–732. doi: 10.1002/art.21650
- Issue published online: 28 FEB 2006
- Article first published online: 28 FEB 2006
- Manuscript Accepted: 21 NOV 2005
- Manuscript Received: 7 OCT 2005
To assess the effects of B lymphocyte depletion on serum B lymphocyte stimulator (BLyS; trademark of Human Genome Sciences, Rockville, MD) levels in patients with rheumatoid arthritis (RA), and to assess the relationship of serum BLyS levels with peripheral blood B cell depletion, levels of autoantibodies and antimicrobial antibodies, the return of peripheral blood B cells, and clinical relapse.
Fifteen patients with active RA underwent rituximab-based B cell depletion therapy (BCDT). Disease activity was assessed clinically, peripheral blood CD19+ B cell counts were determined by flow cytometry, and serum levels of BLyS, IgM, IgA, and IgG rheumatoid factors (RFs), anti–cyclic citrullinated peptide, and antimicrobial antibodies were assessed using enzyme-linked immunosorbent assays.
Peripheral blood B cell depletion was achieved in all 15 patients, and an American College of Rheumatology 20% response was achieved in 13 patients. Following clinical relapse, 7 patients underwent at least 1 additional cycle of BCDT. In every case, serum BLyS levels markedly rose post-BCDT and remained elevated for at least 1–2 months. Serum levels of RF, but not those of anti–tetanus toxoid or anti–pneumococcal polysaccharide antibodies, fell significantly. A decline in serum BLyS levels was associated with the reemergence of B cells in peripheral blood, which, in turn, antedated clinical relapse by variable periods of time. The patterns of B cell depletion, serum BLyS and antibody levels, and clinical relapse for each BCDT cycle were remarkably similar in re-treated patients.
Rituximab-based BCDT leads to marked increases in serum BLyS levels. This may contribute significantly to the survival and/or regeneration of B cell populations capable of triggering clinical relapse.
B cell–targeted therapy promises to be a major avenue for the treatment of a variety of autoimmune diseases (1). Depletion of peripheral blood B cells induced by rituximab, a chimeric anti-CD20 monoclonal antibody (mAb), is clinically effective in patients with rheumatoid arthritis (RA) (2). Among patients achieving disease remission, there is great variability in the timing of clinical relapse. Relapse is nearly coincident with the reappearance of peripheral blood B cells in some patients but in other patients is delayed by up to 3 years after the reappearance of peripheral blood B cells (3). Although previous studies have shown that clinical relapses in patients with RA are associated more closely with increases in circulating autoantibody levels than with reemergence of B cells in peripheral blood (3), the critical sequence of events from clinical response to clinical relapse remains largely unexplored.
Central to an understanding of the mechanism of action of B cell–targeted therapies is a knowledge of the factors controlling B cell survival, expansion, and development. One such key factor is B lymphocyte stimulator (BLyS; trademark of Human Genome Sciences, Rockville, MD [also commonly known as BAFF]), a 285–amino acid member of the tumor necrosis factor (TNF) ligand superfamily (4–6). B cell maturation in BLyS-deficient mice is largely arrested at the T1 maturational stage, and these mice manifest reduced baseline serum immunoglobulin levels and reduced immunoglobulin responses to T cell–dependent and T cell–independent antigens (7, 8). Conversely, administration of exogenous BLyS to mice at the time of immunization with antigen results in enhanced antigen-specific antibody production (9). This is attributable, at least in considerable part, to BLyS-mediated inhibition of B cell apoptosis (9–15). Moreover, repeated administration of BLyS to mice, even in the absence of intentional antigenic immunization, results in B cell expansion and polyclonal hypergammaglobulinemia (4).
Overexpression of BLyS has been compellingly linked to clinical disease in both mice and humans. Circulating autoantibodies characteristic of systemic lupus erythematosus (SLE) (e.g., anti–double-stranded DNA antibodies) and RA (e.g., rheumatoid factor [RF]) often develop in BLyS-transgenic mice as they age (6, 16, 17). Immune complex–mediated glomerulonephritis develops in some of these mice, and many of the mice that do not succumb to renal disease develop features of Sjögren's syndrome (SS) (18). Moreover, SLE-prone (NZB × NZW)F1 and MRL-lpr/lpr mice show decreased disease progression and improved survival following administration of BLyS antagonists (6, 19), and the ongoing joint inflammation and resultant joint destruction associated with collagen-induced arthritis can be inhibited by the BLyS/APRIL antagonist, TACI-Ig (7, 20). In humans, cross-sectional and longitudinal studies have documented BLyS overexpression in as many as 50% of patients with SLE (21–23), and elevated BLyS levels are frequently observed in the sera of patients with RA or SS or in the synovial fluid of patients with RA (18, 21, 22, 24, 25).
Because raised levels of BLyS may promote B cell survival, such BLyS-mediated survival signals might compromise the responses of autoreactive B cells to censoring death signals in patients with autoimmune diseases. Given that the survival of some autoreactive B cell clones may be dependent on levels of BLyS that are higher than those required by nonautoreactive cells (26, 27), alterations in BLyS levels following B cell depletion therapy (BCDT) may profoundly affect the reemergence and/or reexpansion of autoreactive B cells and the recrudescence of autoimmunity and clinical disease. In this report, we describe changes in circulating BLyS levels following BCDT in 15 patients with RA (7 of whom were re-treated with at least 1 further course of BCDT) and the relationship of circulating BLyS levels with peripheral blood B cell depletion and recovery, circulating levels of autoantibodies and antimicrobial antibodies, and the onset of clinical relapse.
PATIENTS AND METHODS
Fifteen patients with clinically active RA received 2 weekly courses of intravenous rituximab (500 mg/dose or 1,000 mg/dose), with or without intravenous cyclophosphamide (750 mg/dose) and/or prednisolone (200–1,425 mg over 3 weeks). The different therapies received by patients, the number of months that passed before the return of B cells, and the number of months that passed before relapse occurred are shown in Table 1. Disease-modifying antirheumatic drugs were discontinued beginning on day 0 (the day on which the first rituximab infusion was administered). Nonsteroidal antiinflammatory drugs and analgesics were administered as needed. Seven patients received a second course of treatment, and 1 patient received a third course.
|Patient/sex/age||Treatment protocol||B cell return, months||Relapse, months||Interval, months, between B cell return and relapse||Response pattern|
|Rx-1||Prednisolone, total mg†||Rx-2||Rx-1||Rx-2||Rx-3||Rx-1||Rx-2||Rx-3||Rx-1||Rx-2||Rx-3|
Clinical evaluations were performed, and peripheral blood total lymphocyte and CD19+ B cell counts were obtained prior to BCDT, at least twice during the first 6 months after BCDT, and, when possible, every 2–3 months thereafter. Serum samples were collected and were stored at −80°C until tested for levels of BLyS, autoantibodies, and antimicrobial antibodies (see below). A positive clinical response was defined as having occurred when the patient achieved a clinical response of ≥20%, according to the American College of Rheumatology (ACR) criteria for improvement in RA (28), at 3 months post-BCDT. Clinical relapse was deemed to have occurred when the ACR-defined clinical response fell below 20% improvement from baseline. Changes in C-reactive protein (CRP) levels were independently used as an objective biochemical measure of disease remission and relapse. This study was approved by the local hospital ethics committee, and all patients gave informed consent.
Assessment of B cell depletion and B cell return.
The normal range for peripheral blood CD19+ B cells, according to the Central Pathology Laboratory at University College London Hospital, is 0.03–0.40 × 109 cells/liter. Depletion of B cells in the peripheral blood was deemed to have occurred when the number of CD19+ B cells was <0.005 × 109/liter, and is reported as 0. In all patients, total peripheral blood B cell depletion (peripheral blood CD19+ B cell count reported as 0) was achieved for at least 5 months. “B cell return” was defined as having occurred when B cells were again detectable in the peripheral blood (i.e., when the CD19+ B cell count was >0.005 × 109/liter). Although in this study B cell return was defined as the time when B cells were again reported as being present in the peripheral blood, the time to full B cell repopulation (i.e., B cell numbers within the normal range) varied considerably between patients and ranged from weeks to several months.
Serum levels of autoantibodies.
Serum levels of IgM-RF, IgG-RF, and IgA-RF were measured by enzyme-linked immunosorbent assay (ELISA), using a commercially available RF detection kit (catalog no. EL-RF/3; TheraTest Laboratories, Chicago, IL) based on the binding of RF to rabbit IgG. For IgG-RF determination, samples were pepsin-digested to avoid interference from IgM-RF and IgA-RF (29). The completeness of pepsin digestion was checked by confirming the absence of binding of horseradish peroxidase–conjugated anti-IgM antibodies to pepsin-digested samples and controls. Normal serum levels for IgM-RF, IgG-RF, and IgA-RF are <25 IU/ml, <20 IU/ml, and <35 IU/ml, respectively. IgG anti–cyclic citrullinated peptide (anti-CCP) antibodies were measured by ELISA (Immunoscan RA; Euro-Diagnostica, Arnhem, The Netherlands), and results are expressed as unit equivalents to the standard serum (normal <200 units/ml).
Serum levels of antimicrobial antibodies.
Serum levels of IgG anti–tetanus toxoid (anti-TT) and anti–pneumococcal capsular polysaccharide (anti-PCP [PCP is a combination of 23 common serotypes]) antibodies were measured by ELISA (Binding Site Limited, Birmingham, UK). Anti-TT antibody levels >0.1 IU/ml were considered to be protective (30). Anti-PCP levels >35 μg/ml are present in 90% of the general population (31).
Serum levels of BLyS.
All analyses were performed using SigmaStat software (SPSS, Chicago, IL). Parametric testing between 2 matched or unmatched groups was performed by paired or unpaired t-test, respectively. When the data were not normally distributed or the equal variance test was not satisfied, nonparametric testing was performed by the Mann-Whitney rank sum test or Wilcoxon's signed rank test, respectively. Correlations were determined by Pearson's product moment correlation for normally distributed data or by Spearman's rank order correlation for data that did not follow a normal distribution.
Serum BLyS and antibody levels following BCDT.
Prior to BCDT, BLyS was not detectable in sera from 9 of the 15 patients with RA (Figure 1A). There were no correlations between pretreatment serum levels of BLyS and pretreatment serum levels of IgA-RF, IgM-RF, and IgG-RF or pretreatment peripheral blood B cell counts (data not shown). By 1–2 months post-BCDT, serum BLyS levels increased in all patients (P < 0.001) (Figure 1A). By 3–4 months post-BCDT, serum BLyS levels, in aggregate, remained elevated compared with baseline levels (P < 0.001) and were unchanged from levels at 1–2 months post-BCDT (P = 0.693). Despite the pronounced changes in serum BLyS levels, changes in serum levels of anti-PCP and anti-TT antibodies were very modest (Figures 1B and C). Pretreatment, 1 patient had a subprotective level of anti-PCP antibodies, and in 4 patients anti-TT antibodies were undetectable. Other than a decline in anti-TT antibody levels to a subprotective level in 1 additional patient at 3–4 months post-BCDT, antimicrobial antibody levels remained in the protective range. In aggregate, anti-PCP antibody levels were unchanged at 1–2 months and 3–4 months post-BCDT compared with pretreatment levels (P = 0.370 and P = 0.224, respectively). Although anti-TT antibody levels declined modestly at 1–2 months post-BCDT (P = 0.026), levels at 3–4 months post-BCDT were restored to baseline levels (P = 0.320 versus pretreatment levels).
In contrast to the limited changes in antimicrobial antibody levels, serum levels of IgM-RF, IgG-RF, and IgA-RF each fell by 1–2 months post-BCDT (P < 0.001, P = 0.004, and P = 0.013, respectively) (Figures 1D–F). By 3–4 months post-BCDT, serum levels of the respective RFs, in aggregate, had declined even further (P = 0.017, P = 0.024, and P = 0.029, respectively, versus levels at 1–2 months post-BCDT).
Pretreatment serum anti-CCP antibody levels were elevated (>200 IU/ml) in 7 of the 10 patients tested (Figure 1G). Although they visually paralleled the changes in RF levels, the changes in serum anti-CCP antibody levels at 1–2 months and 3–4 months relative to pre-BCDT levels did not quite achieve statistical significance (P = 0.074 and P = 0.098, respectively).
Patterns of B cell return and onset of relapse in patients receiving 1 or more courses of BCDT.
As shown in Table 1, 13 of the 15 patients in this study had a positive clinical response (20% improvement or better according to the ACR criteria). All patients experienced clinical relapses, and 7 of them were re-treated with rituximab-based BCDT. The 13 responding patients were stratified into 2 groups based on the temporal relationship between reemergence of B cells (>0.005 × 109 CD19+ B cells/liter) in the peripheral blood and development of clinical relapse. In 7 patients (group A), there was at least a 5-month interval between the time of return of peripheral blood B cells and clinical relapse (median 11 months). In the other 6 patients (group B), clinical relapse ensued no later (and often sooner) than 2 months after the return of B cells in the peripheral blood. The pattern for any given patient (response pattern A or B) following re-treatment was identical to that following the initial course of BCDT (Table 1).
Individual patients varied in the manner in which B cell recovery occurred after B cells first reappeared in the peripheral blood (i.e., when the CD19+ B cell count was >0). Some patients experienced rapid B cell repopulation, with peripheral blood B cell counts reaching the normal range within 4–8 weeks, whereas B cell numbers rose more slowly (i.e., over the course of several months) in other patients. The distribution of patients in whom repopulation in the periphery occurred rapidly and patients in whom repopulation of B cells to the normal range took longer was equal in groups A and B (data not shown).
Relationship between serum BLyS levels, B cell depletion, and B cell return.
BLyS levels rose following BCDT and, at 3–4 months posttreatment (when all patients were still experiencing B cell depletion), BLyS levels remained elevated (Figure 1A). Thus, a strong inverse relationship between serum BLyS levels and depletion of B cells from the peripheral blood was observed. To illustrate whether this relationship was reversed upon the return of B cells, the results of serial studies of BLyS levels and CD19+ B cell counts in 2 patients, representative of the 2 response patterns, are shown in Figure 2A (group A) and Figure 2B (group B). B cell numbers in both of these patients reached the normal range within weeks of the first recorded positive CD19 cell count post-BCDT, but the patients experienced relapse at different times. Similar patterns and timing of B cell return and repopulation and fluctuations in BLyS levels were observed in these and the other re-treated patients (Table 1, and data not shown).
In the group B patient, the drop in the BLyS level to its pretreatment value was very closely associated with increasing numbers of peripheral blood B cells (Figure 2B). Results for the group A patient (Figure 2A) showed a rapid rise in peripheral blood B cell numbers to above the lower limit for normal within 1 month after their initial detection post-BCDT, followed by a more gradual increase. Although BLyS levels did decrease toward pretreatment levels in this patient following B cell return, the decrease was more gradual and slower than that in the group B patient. Regression analysis of B cell numbers and BLyS levels, which were followed up until clinical relapse 17 months after B cell return, revealed a significant inverse correlation (R = 0.80, P = 0.003) in this patient. However, similar analyses of B cell numbers versus BLyS levels in the other 4 group A patients from whom sufficient samples were available for analysis revealed a significant (P < 0.05) inverse relationship in only 2 of these individuals. In addition, the presence of raised pretreatment serum BLyS levels did not correlate with the time required for B cells to return to the peripheral blood (P = 0.079) (data not shown). These results suggest that the relationship between the timing and speed of B cell return and BLyS levels is likely to be complex.
Serologic parameters and BLyS levels.
The relationships between BLyS levels, CRP levels, circulating antibodies, and clinical relapse are shown for 2 representative patients from each group in Figures 3 and 4. In all 4 of these patients, serum BLyS levels rose, and serum IgM-RF levels fell, with the latter reaching nadirs prior to peripheral blood B cell return. In group A patients (Figure 3), B cells reemerged in the peripheral blood months before a rise in the CRP level and clinical relapse. IgM-RF levels rose before or at the time of relapse and the rise in the CRP level. In group B patients (Figure 4), reemergence of peripheral blood B cells was temporally closely associated with the onset of clinical relapse after each course of BCDT. As was the case for group A patients, a rise in serum IgM-RF levels preceded clinical relapse in group B patients. Importantly, serum IgM-RF levels began to rise in patient 2 (Figures 4C and D) even while the peripheral blood was depleted of B cells and serum BLyS levels were still elevated compared with pretreatment levels. Changes in serum anti-PCP antibody levels varied greatly among patients but generally did not mirror changes in serum RF levels, indicating that the fall and rise of RF levels were not merely reflections of global changes in circulating antibody levels. No relationship between clinical relapse and anti-PCP antibody levels was noted. Similar patterns of serologic responses as observed after initial treatment were seen following subsequent treatment courses in patient 7 (Figures 3A and B) and patient 2 (Figures 4C and D).
Relationship between serum BLyS levels and timing of clinical relapse.
Among the 7 group A patients, in whom clinical relapse was temporally removed (range 5–37 months) from peripheral blood B cell return, pretreatment and 1–2-month post-BCDT serum BLyS levels were no different from those in the group B patients (n = 6), in whom clinical relapse was temporally associated with reemergence of B cells in the peripheral blood (P = 0.445 and P = 0.453, respectively) (data not shown). There were also no differences in serum BLyS levels between the 2 groups of patients 3–4 months following BCDT (P = 0.588) (Figure 5). However, there was a striking difference in BLyS levels between group A patients and group B patients at the time of B cell return to the peripheral blood (Figure 5). Median BLyS levels were significantly higher in the later-relapsing group A patients (6.04 ng/ml) compared with the group B patients (1.16 ng/ml), in whom B cell return was closely associated with clinical relapse (P = 0.018). At the point in time when group A patients did experience clinical relapse, there was no significant difference in serum BLyS levels between the 2 groups (P = 0.299).
Following depletion of B cells from the peripheral blood of rituximab-treated patients with RA, serum BLyS levels rose markedly and remained elevated for varying periods of time in all patients (Figure 1A). In patients treated more than once, similar rises in serum BLyS levels occurred following each course of BCDT (Figure 2). Because the expression of BLyS receptors is largely restricted to B cells (4, 33–37), the increase in serum BLyS levels following BCDT may be attributable, at least in part, to the physical loss of BLyS-binding B cells (26). Alternatively, BLyS production may be under homeostatic control from an as-yet-unidentified negative feedback signal from the B cell compartment that is deficient during periods of B cell depletion. Several cytokines, including interferon-α (IFNα), IFNγ, and interleukin-10 (IL-10), can up-regulate BLyS messenger RNA levels in, and BLyS protein production by, myeloid-lineage cells (e.g., monocytes, macrophages, dendritic cells, neutrophils) (32, 38, 39). In addition, certain cytokines, such as IL-8 or TNFα, can promote the release of intracellular BLyS stores from neutrophils (40). Although these cytokines may be implicated in the elevated circulating BLyS levels associated with periods of disease activity, it is uncertain what role these cytokines may play during the periods of clinical remission, accompanied by falling levels of acute-phase proteins, following BCDT.
The increases in serum BLyS levels during the first 4 months post-BCDT were anti-parallel to the concomitant decreases in serum levels of IgM-RF, IgG-RF, and IgA-RF (Figures 1D–F). The substantial and sustained reductions of RF levels beyond their molecular half-lives suggest that the B cell precursors of RF-producing plasma cells were sensitive to the lytic properties of rituximab. In addition, the considerable lengths of time between BCDT and clinical relapse (invariably associated with rises in serum RF levels) suggest that the RF-producing plasma cells were not immediately replenished from a persisting memory B cell population.
In contrast to the anti-parallel relationship between serum levels of BLyS and RF, levels of antimicrobial antibodies remained stable following BCDT (Figures 1B and C). Indeed, BCDT-induced increases in serum BLyS levels were followed, after a short delay, by increases in serum anti-PCP antibody levels in 6 of the 15 patients (Figure 3B, Figure 4D, and results not shown). Given that normal (nonmalignant) plasma cells do not express surface CD20 and, therefore, are not susceptible to rituximab-induced death, the rise in serum anti-PCP antibody levels following BCDT raises the possibility that the proportion of nonautoreactive antibody-secreting cells (e.g., those secreting anti-PCP antibodies) that are fully differentiated to the CD20− plasma cell stage is greater than the proportion of autoreactive antibody-secreting cells (e.g., those secreting RF autoantibodies). Rituximab would preferentially delete autoreactive B cells, compared with anti-PCP–committed B cells, and the concomitant elevated serum BLyS levels would promote increased endogenous production of nonautoreactive antibodies due to their expansion into the space vacated by the deleted autoreactive B cells.
Despite the fact that the serum half-life of rituximab is only ∼20 days (41), and despite the development of high serum BLyS levels soon after BCDT, B cells did not reemerge in the peripheral blood of our patients for 5–14 months (Table 1). Very little is known concerning the triggers for restoration of B cells in peripheral blood following their depletion. We observed no correlation between BLyS levels pre-BCDT and the length of time required for reemergence of B cells in the periphery. There may, however, be important differences between individual patients in the degree of noncirculating B cell expansion that occurs before B cells reenter the circulation. Despite an ample supply of BLyS, protracted periods of time may be required to reengage the pathways leading to generation of a pool of memory B cells sufficiently large to trigger relapse. In later-relapsing patients (such as group A in our study), B cells may reemerge in peripheral blood in conjunction with relatively little B cell expansion in secondary lymphoid tissues. Thus, the total B cell load in such patients would be relatively low at the time of B cell return to the peripheral blood; therefore, serum BLyS levels would remain relatively high. With continued stimulation from BLyS, the total load of pathogenic B cells would steadily increase over time and finally reach the requisite threshold for clinical relapse. In contrast, B cells in patients such as those in group B may return to the peripheral blood at a time when memory B cells have already undergone considerable expansion in secondary lymphoid tissues. At this point, the total load of pathogenic B cells would already be at or near the threshold for promoting clinical relapse, and serum BLyS levels would therefore not be as high as those in later-relapsing (group A) patients.
Alternatively, there may be important differences among patients in terms of the percentages of remaining pathogenic cells within the B cell compartment. For instance, the percentage of pathogenic B cells in later-relapsing (group A) patients may be relatively low. Even at the time that B cells return to the peripheral blood, the total load of pathogenic B cells may still be insufficient to drive clinical relapse. It is possible that without elevated serum BLyS levels, the pathogenic B cells in these patients would never reexpand to the point of promoting clinical relapse. Conversely, the percentages of pathogenic B cells in group B–type patients may be relatively high. By the time B cells reemerge in peripheral blood, the total load of these pathogenic cells is sufficiently great to trigger clinical relapse.
The fact that all patients do ultimately experience relapse implies that during the period of B cell depletion from the peripheral blood and amelioration of clinical disease activity, the underlying disease process, nonetheless, still persists in a dormant state capable of eventual reactivation. In patients with lymphoma, the effectiveness of rituximab-based depletion of B cells is linked to genetic polymorphisms of Fc receptors and complement-regulatory proteins (42, 43). In patients with SLE, the Fcγ receptor III genotype dictates the levels of peripheral blood B cell depletion when escalating doses of rituximab are used (44). Thus, clinical relapse in rituximab-treated patients with RA may be a direct consequence of the survival of some pathogenic memory B cells that escaped depletion or may be attributable to long-lived plasma cells. In addition, the persistence of autoreactive T cells in rituximab-treated patients could lead to activation of newly generated pathogenic B cell clones bearing specificities different from those of the “original” pretreatment pathogenic B cells.
Our findings suggest that the beneficial effects of rituximab-based BCDT may be offset by an associated rise in circulating BLyS levels, thereby promoting support for surviving or reemerging B cells involved in disease pathogenesis. In human CD20–expressing mice, combination treatment with anti-CD20 mAb and a BLyS antagonist depleted virtually all B cells (45). Accordingly, neutralization of BLyS as an adjunct to rituximab therapy in human RA patients may have the potential to extend the duration of clinical remission. Clinical trials to directly test this hypothesis appear to be warranted.
- 30The immunological basis for immunization series. Module 3: tetanus. Geneva: World Health Organization; 1993..