Safety of Herpes Zoster Vaccine in the Shingles Prevention Study

Herpes zoster occurs with increasing frequency and severity with increasing age (1, 2). It is often associated with pain and discomfort that may interfere with functional status and quality of life during the acute phase. Herpes zoster pain and discomfort may persist for weeks, months, or even years. This debilitating complication, known as postherpetic neuralgia, results in significant decrements in quality of life and ability to perform activities of daily living (2–4). Antiviral therapy has limited effect on the frequency and severity of postherpetic neuralgia. Therefore, a safe and effective vaccine to prevent herpes zoster and postherpetic neuralgia in older adults at greatest risk is highly desirable.

We and others (5, 6) have reported that live attenuated Oka/Merck herpes zoster vaccine (Merck & Co., Whitehouse Station, New Jersey) is immunogenic in populations who have had varicella-zoster virus (VZV) infection, including older adults, and in persons lacking VZV antibody (6). Veterans Affairs Cooperative Study 403 (SPS [Shingles Prevention Study]) (7, 8) showed that herpes zoster vaccine was effective in preventing herpes zoster and postherpetic neuralgia in persons 60 years or older. Health care providers and patients need detailed information about the safety and tolerability, as well as efficacy, of a new vaccine to make informed decisions about its use. Determining the safety profile of herpes zoster vaccine was a major study objective. We present a comprehensive analysis of our observations.

Methods

Design Overview

The methods have been published elsewhere (7). The study was a randomized, placebo-controlled trial of herpes zoster vaccine designed to test the vaccine's safety and efficacy. All participants and follow-up study personnel, except for personnel administering the vaccine, were blinded to treatment throughout the study until data were reviewed for accuracy and the database was locked. The study was approved by the Human Rights Committee of the West Haven Department of Veterans Affairs Cooperative Studies Program Coordinating Center, West Haven, Connecticut, and by the institutional review board at each site.

Setting and Participants

The study was conducted at 22 academic medical centers in the United States between 1998 and 2004. We enrolled immunocompetent adults 60 years or older with a history of chickenpox or more than 30 years of residence in the continental United States and no history of herpes zoster. The mean follow-up was 3.4 years.

Randomization and Intervention

At each site, we stratified consenting eligible participants by age (60 to 69 years or ≥70 years) and randomly assigned them to receive investigational herpes zoster vaccine or placebo. The Coordinating Center generated a permutated block randomization. Allocation to herpes zoster vaccine and placebo was balanced in blocks of 6 vials and in both of the prespecified age strata for each of the 22 study sites. Single-dose vials of herpes zoster vaccine (median potency, 24 600 plaque-forming units per dose) and placebo were maintained at −20 °C. Vials were reconstituted, and participants were inoculated subcutaneously in the nondominant deltoid region within 30 minutes after the vial was removed from the freezer. Study personnel who reconstituted the vials and inoculated the participants had no subsequent contact with them and no subsequent role in data collection or analysis.

Follow-up Procedures and Outcome Measures

Participants and study personnel responsible for follow-up assessments were blinded to treatment assignment. Participants were followed monthly, either by participant-initiated reports to a toll-free automated telephone response system or by direct calls from study site personnel. In addition, all participants were questioned about the occurrence of rashes and serious adverse events during the 42 days after inoculation.

Early in the enrollment phase, approximately 300 participants at each study site (6616 overall) were also enrolled in an adverse events substudy. Participation in the substudy was voluntary and not randomized. Participants were asked to complete a detailed vaccination report card designed to capture all adverse events that occurred during the first 42 days after inoculation. The vaccination report card prompted persons to record oral temperature for the first 21 days, erythema, swelling, pain and tenderness, and rash at the inoculation site; any other inoculation-site–related signs or symptoms; rashes away from the inoculation site; exacerbations of preexisting diseases; new local or systemic illnesses; hospitalizations; and any other event the patient considered medically important.

Serious adverse events, defined by concurrent U.S. Food and Drug Administration and International Committee on Harmonization guidelines (9), were monitored throughout the study by both active and passive surveillance. They were reported on study-specific MedWatch forms. During the first 42 days after inoculation, all enrolled persons were followed actively for serious adverse events. After that, investigators were instructed to report all deaths as well as any serious adverse event considered to be possibly, probably, or definitely related to vaccination. In addition, persons in the substudy were questioned monthly regarding hospitalizations. All MedWatch forms were reviewed by the Cooperative Studies Program Research Pharmacist, the National Study Coordinator, the Study Chairman, SPS personnel at the Coordinating Center, and personnel at Merck (investigational new drug holder for the vaccine). We resolved any questions by querying the site that reported the serious adverse event.

Each adverse event was recorded on a specific study form and, before the study was unblinded, was coded by the site by using a controlled vocabulary system (Coding Symbols from a Thesaurus of Adverse Reaction Terms [COSTART]) (10). Three investigators blindly reviewed all serious adverse events reported during the 42 days after inoculation and assigned each event to 1 of 6 pathophysiologic categories. Appendix Table 1 defines these categories. Discrepancies in assignment were resolved by discussion among these investigators.

Statistical Analysis

All data were stored and analyzed at the Coordinating Center. We performed analyses by using SAS statistical software (SAS Institute, Cary, North Carolina). We calculated the rates of adverse events during the first 42 days after inoculation by dividing the number of persons with the event by the number of persons with safety follow-up. We calculated risk differences between groups (vaccine − placebo) and 95% CIs by using an asymptotic method for the difference between 2 proportions; for analyses of treatment differences, including both age cohorts, we weighted the proportions by the number of participants with safety follow-up in each age stratum at the 22 sites (11). We calculated rates of serious adverse events, hospitalization, or death occurring at any time during follow-up by dividing the number of persons with at least 1 event by the number of person-years of safety follow-up. Cumulative event rates were calculated by using product-limit estimates for time-to-event data and compared treatment groups and age strata by using a log-rank test stratified by site. We performed post hoc analyses of severity and duration in participants who had an adverse event. Duration of adverse events was compared between groups by using the Wilcoxon rank-sum test, and severity of adverse events was compared between groups by using the Cochran–Mantel–Haenszel chi-square test statistic stratified by age and site. We used the time to the first occurrence of a serious adverse event in each person for the time-to-event analyses. When assessing severity of inoculation site or systemic adverse events, we used the report of the most severe adverse event per person, rather than the first report. We did not prespecify within-age stratum comparisons of risk between treatments; however, we performed them in response to questions to the study group on the safety of the vaccine in more elderly persons.

Role of the Funding Source

The trial was funded by the Cooperative Studies Program of the Department of Veterans Affairs, Office of Research and Development, and by a grant to the Veterans Affairs Cooperative Studies Program from Merck. Merck was involved in review of all completed MedWatch forms but not in coding the adverse events, contacting the sites, performing the statistical analyses reported here, preparing the manuscript, or submitting the manuscript for publication. An executive committee that included 2 nonvoting members employed by Merck was primarily responsible for the conduct of the study.

Results

A total of 38 546 older adults participated in this study (Figure 1). During the first 42 days after vaccination, we obtained safety follow-up data from 37 388 (97%) participants. A total of 6575 (99%) of the 6616 substudy participants completed the vaccination report card. More than 95% of the 38 546 trial participants were followed to the end of the study and completed a closeout interview (mean safety follow-up, 3.39 years; range, 1 day to 5.40 years). In persons without closeout interviews, 1598 (4.1%) were known to have died, 132 (0.3%) withdrew, and 113 (0.3%) were lost to follow-up. The proportion of persons completing safety surveillance and the reasons for or rates of incomplete follow-up did not differ between groups.

• Download figure
• Download PowerPoint

In all participants, the frequency and distribution of vesicular rashes occurring during the first 42 days after vaccination differed between vaccine and placebo recipients. A varicella-like rash, defined as 1 or more ungrouped vesicles, occurred at the inoculation site more frequently in vaccine recipients than placebo recipients (0.11% vs. 0.04%); however, rates of rashes occurring elsewhere were similar (0.10% vs. 0.07%). The varicella-like rashes occurring at the inoculation site typically consisted of small numbers of vesicles, did not spread, and were transient (mean duration, 5.4 days in vaccine recipients and 6.7 days in placebo recipients). Varicella-like rashes away from the inoculation site tended to last longer than inoculation-site rashes (mean duration, 17.6 days [range, 3 to 110 days] in vaccine recipients and 18.6 days [range, 1 to 92 days] in placebo recipients). Of the non–herpes zoster rashes, 11 were evaluated by central polymerase chain reaction (PCR) assay, and 5 were evaluated by local virus culture. All results were negative for both wild-type and Oka vaccine strain VZV. One vaccine recipient had a disseminated varicella-like rash on day 18 that was assessed as probably related to the inoculation but was not suspected to be herpes zoster. During the 42 days after inoculation, herpes zoster–like rashes (defined as multiple vesicles in a dermatomal distribution) occurred more often in placebo recipients than vaccine recipients. Of these, herpes zoster was confirmed in 24 placebo recipients and 7 vaccine recipients. Specimens were available from all but 1 case in each treatment group, and all 29 tested positive for wild-type VZV DNA by PCR assay. None tested positive for the Oka vaccine strain VZV.

In substudy participants, the most common adverse events at the inoculation site were erythema, swelling, and pain and tenderness (Table 1). For both treatment groups, inoculation-site adverse events were more common in younger persons than in older persons (Table 1). Most events were mild or moderate in severity (Table 2). Erythema and swelling at the inoculation site were the only events in which the intensity was statistically significantly greater in vaccine recipients than in placebo recipients. However, fewer than 1% of vaccine recipients reported them as severe. Erythema, swelling, pain and tenderness, and warmth at the inoculation site persisted for longer in vaccine recipients than placebo recipients (Appendix Table 2). Erythema was the only local adverse event that persisted longer in younger vaccine recipients than older recipients (data not shown). Pruritus at the inoculation site occurred more frequently in vaccine recipients than placebo recipients in both age strata (Table 1) and persisted longer in vaccine recipients than in placebo recipients (Appendix Table 2). Inoculation-site events occurred sooner after vaccination in placebo recipients than vaccine recipients (mean, 1.7 vs. 2.3 days; P < 0.001), whereas the time to the first systemic adverse event (mean, 15.9 vs. 15.2 days) and the duration of systemic adverse events (mean, 14.8 vs. 19.1 days) were similar in vaccine and placebo recipients (data not shown).

During the first 42 days after inoculation, serious adverse events were reported in 1.4% of all participants in each group (Table 3). The point estimates for risk difference between treatment groups were 0.12% or less when analyzed overall, by prespecified age strata, or by COSTART body system (Table 3). Rates of serious adverse events increased with age at similar rates in vaccine recipients and placebo recipients. A time-to-event analysis by age group for the whole study population showed no statistically significant difference between vaccine recipients and placebo recipients (Figure 2).

• Download figure
• Download PowerPoint

We performed a post hoc analysis for serious adverse events that occurred during the 42 days after inoculation in all participants 80 years or older (Table 3). Follow-up of serious adverse events was available for 96.6% (1220 of 1263) of vaccine recipients and 96.6% (1289 of 1335) of placebo recipients in this age group. The overall rate of serious adverse events did not statistically significantly differ by treatment group in participants 80 years or older (risk difference, 0.6 percentage points [95% CI, −0.5 percentage points to 1.7 percentage points]), and there were no statistically significant differences between groups for any body system (COSTART) or Physiologic Diagnostic Category classification (data not shown).

In substudy participants, the overall rate of serious adverse events during the 42 days after inoculation was 1.6% (Table 4). As reported elsewhere (7), the rate of serious adverse events was higher in vaccine recipients than in placebo recipients (1.9% vs. 1.3%; P = 0.038, analysis stratified by age and site). Although some serious adverse events occurred more often in vaccine recipients than in placebo recipients (for example, cardiovascular body system classification in 20 vs. 12 persons, respectively), none of the differences at or below the level of body system was statistically significant (7). Because of limitations in diagnostic classifications based on body system, we further analyzed all serious adverse events that occurred during the first 42 days after vaccination by using pathophysiologic criteria (Appendix Table 1). We subclassified serious adverse events related to vascular disease on the basis of whether they were consistent with acute vascular pathology (for example, myocardial infarction, strokes) or with functional disturbance associated with underlying vascular disease (for example, congestive heart failure). The overall rates of vascular events (that is, vascular [pathology] plus vascular [functional]) were nearly identical between vaccine recipients and placebo recipients in the total study population (99 [0.5%] vs. 101 [0.6%]) (Table 3). When each category of vascular pathophysiology was considered separately, there were small numerical differences between treatment groups that were neither statistically significant nor clinically meaningful (Table 3). We observed similar results when we analyzed serious adverse events in the substudy (Table 4).

Over the course of the entire study, rates of death in the total SPS population (Appendix Figure 1) and rates of hospitalization in the substudy (Appendix Figure 2) were greater in the older age stratum than in the younger age stratum. However, rates for each treatment group, both overall and by age strata, were essentially identical and did not vary appreciably over the course of the study. Of note, the death rate in study participants was substantially lower than that reported by the U.S. National Health Statistics for the general U.S. population 60 years or older (12). When we stratified the study population into 5-year age groups to match the U.S. National Health Statistics data, the mortality rate in each age group was less than half (43% to 49%) of that reported for the general population. This is predictable given the enrollment procedures, which excluded persons expected to have limited life expectancy or comorbid conditions associated with high risk for herpes zoster complications (for example, known immunosuppression), to ensure the safety of study participants and adequate duration of follow-up.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Discussion

As shown elsewhere by the SPS (7), 1 dose of herpes zoster vaccine reduced the burden of illness due to herpes zoster, as well as the incidence of postherpetic neuralgia and herpes zoster. Our report further demonstrates that herpes zoster vaccine had remarkably low rates of acute local reactions and, across the study population, had no detectable effect on the rates of serious adverse events during the 42 days after inoculation or on the rates of death during the entire mean 3.39 years of follow-up.

An important safety consideration for any live attenuated virus vaccine is that the vaccine not cause the disease it is designed to prevent. We reviewed all rashes reported by study participants during the 42 days after inoculation. Vesicular rashes at the injection site were more common in vaccine recipients than in placebo recipients, but overall these events were infrequent and limited in extent and duration. Neither wild-type nor Oka vaccine strain VZV was shown in these injection-site lesions, either by culture or PCR assay, and there were no documented episodes of disseminated vesicular disease caused by vaccine virus.

The prespecified primary efficacy analysis excluded cases of herpes zoster that occurred during the first 30 days after vaccination. However, in the safety analysis of all vesicular rashes during the first 42 days after inoculation, herpes zoster occurred more frequently in placebo recipients than in vaccine recipients (24 vs. 7 confirmed cases, respectively), indicating not only that herpes zoster vaccine did not cause herpes zoster but that it protected against herpes zoster during this early period, as well as later. The early onset of vaccine-induced protection is consistent with immunologic studies (13) indicating that VZV-seropositive, latently infected persons have an anamnestic response to herpes zoster vaccine.

Inoculation-site reactions were the primary difference in the rate of adverse events in recipients of herpes zoster vaccine versus placebo in the immediate postinoculation period. These reactions were typically transient and rarely severe. Overall, the reactions were similar to those observed in recipients of other vaccines recommended for older adults (14). Inoculation-site reactions were statistically significantly more frequent and intense in younger vaccine recipients than in older recipients. We reported elsewhere (5) that the immune response to herpes zoster vaccine was more vigorous in younger participants than in older participants. These findings suggest that the local side effects of the herpes zoster vaccine may have been mediated by immune responses to the attenuated vaccine virus.

Per protocol, during the 42 days after inoculation, there was active surveillance across the study for the occurrence of serious adverse events, and nearly all persons were explicitly questioned. There were no differences or suggestive trends in the frequency or distribution of serious adverse events between vaccine and placebo recipients in the total study population in any of the prespecified analyses, including by age strata, COSTART body system and subterms, and time-to-event analysis. The last is particularly compelling because vaccine-associated serious adverse events would be expected to show temporal clustering (15, 16).

While our study was in progress, reports based on uncontrolled observations (17–20) suggested that smallpox vaccine was associated with acute cardiovascular adverse events, including ischemia. One proposed mechanism (17) was that inflammatory mediators (for example, γ-interferon and tumor necrosis factor) generated during the immune response to that live virus vaccine might have increased the risk for acute vascular pathology. Therefore, in addition to the prespecified safety analyses, we conducted 2 post hoc analyses to address concerns about any possible vaccine-associated increase in the risk for vascular events in older adults, who are the target population for herpes zoster vaccine. Rates of serious adverse events in the 42 days after inoculation did not statistically significantly differ in vaccine and placebo recipients 80 years or older, potentially the most vulnerable trial participants. In addition, while blinded to individual treatment assignments, we classified each adverse event on the basis of inferred pathophysiology, with particular emphasis on cardiovascular and cerebrovascular events. These analyses also found no statistically significant or clinically meaningful differences between recipients of herpes zoster vaccine and placebo.

The substudy, which included approximately one sixth of the total study population, was designed to provide detailed information about relatively high-frequency, vaccination-related events (for example, inoculation-site events). Although participants were not selected at random, enrollment was well balanced by treatment in each age stratum. The proportion of persons with 1 or more serious adverse events was essentially the same in the substudy as in the total study population, indicating that ascertainment of the low-frequency but medically important events was similar in both populations. However, in contrast to the total study population, serious adverse events in the substudy were statistically more frequent in vaccine recipients than in placebo recipients (P = 0.038). This difference was not reflected in the prespecified analyses by age strata or by using the COSTART classification method; specifically, there were no statistically significant differences for specific events or at the level of body system except for the sight/sense system, in which there were only 4 events. Our post hoc analysis, which was blinded and in which serious adverse events were classified by pathophysiologic category, also revealed no statistically significant differences in rates of these events in the recipients of herpes zoster vaccine versus placebo in the substudy. In addition, beyond 42 days after inoculation, the treatment groups did not differ in any analysis. On the basis of all available data and analyses, we conclude that the observed difference in rates of serious adverse events in the vaccine recipients and placebo recipients in the substudy, although statistically significant, represents a chance occurrence in a selected subgroup and does not reflect vaccine-related events.

Our study has limitations. At enrollment, participants were ambulatory and noninstitutionalized and 95% were white. We excluded severely debilitated older adults and those with known immunosuppressive disease or treatment. The safety and efficacy of herpes zoster vaccine in such populations are uncertain. Participants in the substudy were not randomly selected. Serious adverse events not treated at study sites were assessed by participants' reports and, although efforts were made in every case, these were not always confirmed by medical record review.

In summary, our analyses showed that herpes zoster vaccine was well tolerated and safe in older immunocompetent adults. There was a modest increase in the rate of acute inoculation-site events in vaccine recipients, but no increased risk for herpes zoster itself and no pattern suggesting any serious adverse events were causally related to vaccination. Given the substantial protection that herpes zoster vaccine provides against the occurrence and morbidity of herpes zoster and, specifically, postherpetic neuralgia, we believe that this safety profile supports the recommendation for routine use of herpes zoster vaccine in immunocompetent older adults, who are at increased risk for herpes zoster and its complications (21).