Evaluation of the Safety and Immunogenicity of a Booster (Third) Dose of Inactivated Subvirion H5N1 Influenza Vaccine in Humans

Direct transmission of highly pathogenic avian influenza viruses of the H5N1 subtype continues to occur, human-to-human transmission has been noted, and international organizations have identified vaccine development as a critical strategic priority [1, 2]. We previously evaluated a 2-dose regimen of an unadjuvanted subvirion H5N1 vaccine [3] that was subsequently approved by the US Food and Drug Administration (in April 2007), making it the first avian influenza vaccine with this distinction in the United States [4]. Our data and previous work [5], however, indicate that H5-containing vaccines are only modestly immunogenic. Here, we report the safety and immune response after a third dose of this vaccine when given 6 months after the second dose—a strategy to enhance the immunogenicity of H5 vaccine.

Methods. We evaluated the safety and immunogenicity of a third dose of an influenza A/Vietnam/1203/2004(H5N1) vaccine among healthy adults 18–64 years of age who had previously received 2 doses of this vaccine [3]. The prior study was a placebo-controlled, double-blind, multicenter trial that included 451 participants randomized to receive either placebo or vaccine at a dose of 90, 45, 15, or 7.5 µg, administered intramuscularly in a 2-dose schedule separated by 28 days. In the present study, 337 study participants received a third dose of the same vaccine dose they had received in the first study, with prior placebo recipients excluded. The third dose was administered 6 months after the second dose. All participants and study personnel remained blinded to the recipient's vaccine dose, and the vaccinator was not involved in the assessment of adverse events. Participants were observed for 30 min after each dose for immediate adverse events and maintained a memory aid for the next 7 days in which they recorded the presence and severity of local and systemic symptoms and their oral temperature. We used a standardized scale by which symptoms were considered mild if they did not interfere with normal activities, moderate if they resulted in some interference with normal activities, and severe if they prevented normal daily activities.

The memory aid was reviewed on day 7 after vaccination, and medical history and interval adverse events were also reviewed on days 28 and 56. Serum for assessment of antibody responses was obtained before vaccination as well as 28 days and 6 months thereafter. An independent data and safety monitoring board and local independent safety monitors (administered by the National Institutes of Health) were active for this trial. This study was approved by the institutional review boards of the Los Angeles Biomedical Research Institute at the Harbor-UCLA Medical Center, the University of Rochester, and the University of Maryland.

Vaccine was created in the same manner as for our earlier study [3]. The reference reassortant was generated from the human isolate influenza A/Vietnam/1203/2004(H5N1) virus by use of a plasmid-rescue system [6]. The rescued virus derived the gene segments encoding the hemagglutinin (HA) and neuraminidase from the A/Vietnam/2004 virus and all other genes from A/PR/8/34 virus, a laboratory strain commonly used as a base platform for influenza vaccines. The HA gene was modified to replace the nucleotides encoding 6 basic amino acids at the HA1/HA2 cleavage site associated with high pathogenicity in birds with an avirulent avian sequence. The resulting rgA/Vietnam/ 1203/2004×A/PR/8/34(H5N1) virus was antigenically identical to the wild-type A/Vietnam/1203/04 virus. The vaccine product was generated using standard techniques [3] and then stored at 4°C until use.

Microneutralization (MN) and hemagglutination-inhibition (HAI) assays were performed at a central laboratory (SRI) using the rgA/Vietnam/1203/2004×A/PR/8/34(H5N1) vaccinereference reassortant virus.MNassays were performed as previously described [3, 7–10]. Serum samples were tested at an initial dilution of 1:20, and samples that were negative were assigned a titer of 10. HAI assays were performed using established procedures with horse erythrocytes [9, 10]. The serum samples were tested at an initial dilution of 1:20 after treatment with receptordestroying enzyme to remove nonspecific inhibitors of agglutination, and samples that were negative were assigned a titer of 10. For both assays, serum samples were tested in duplicate in independent assays, and serum that gave >2-fold different results in replicate samples were retested.

The prespecified primary immunologic end point for this trial was the proportion of subjects in each group who developed a neutralizing titer against the rgA/Vietnam/1203/04 virus of 1:40 twenty-eight days after the third dose of vaccine. Exact confidence intervals (CIs) are reported for all proportional end points. Geometric mean titers (GMTs) of antibody were compared between groups by the Wilcoxon rank sum test, and response rates were compared between groups by Fisher's exact test and within groups (between doses) by McNemar's test. Generalized linear and logistic regression models were used for evaluating differences in immunologic response due to sex, age, and dose, while controlling for the baseline titer value and the receipt of a prior vaccination.

Logistic regression models were used for evaluating the overall differences in reactogenicity rates, on the basis of the dichotomization into “none or mild” versus “moderate or severe” for the most severe response after vaccination. All reported P values are 2-sided. All data analyses and statistical computations were conducted with SAS (version 9.1). Given a sample size of at least 75 participants per group, the half-width of a 95% CI for any observed event rate is no greater than 12%. In addition, the binomial probability of detecting 2 or more adverse events (within a vaccine-dose group) is at least 89% for true event rates of5%or higher.

Results. We enrolled and vaccinated 337 participants in this trial, all of whom are included in the safety analyses. Two participants were excluded from the immunogenicity analyses when it was determined that they had previously received a different H5 influenza vaccine in another clinical trial. Two others were lost to follow-up after vaccination. Therefore, the immunogenicity analyses included 84, 86, 78, and 85 (333 in total) participants for the 7.5-, 15-, 45-, and 90-µg formulations of vaccine, respectively. The median age was 40 years (range, 19–64 years); 54% were female, 82% white, 11% Asian, 5% African American, and 90% non-Hispanic. Approximately 42%–44% of participants in each dose group had received the standard influenza vaccine in the 2005–2006 season before enrollment. Participant age and racial/ethnic distributions were similar between each of the vaccine-dose concentrations.

The vaccine was generally well tolerated at all dose levels. Two hundred twenty-eight participants (68%) reported reactogenicity symptoms, amongwhom 86% reported only mild symptoms, and no serious allergic-type reactions or serious local reactions occurred.

Pain and local tenderness at the injection site were the most commonly reported adverse events. Moderate or severe pain and/or tenderness was reported in only 1, 2, 2, and 3 participants in the 7.5-, 15-, 45-, and 90-µg-dose groups, respectively (odds ratio [OR], 0.99 [95% CI, 0.97–1.01]). Systemic symptoms were very uncommon and only 3 were graded as severe: malaise and feverishness (45-µg dose), temperature >39°C (45-µg dose), and malaise and feverishness (15-µg dose). Only 4 participants (1.2%) had fever (defined as >37.8°C) after vaccination; 2 after receipt of 15 µg and 2 after receipt of 45 µg. Participants who reported a moderate or severe adverse event after the second dose (n = 29 [8.6%]) were more likely to report such an event after the third dose (OR, 8.8 [95% CI, 3.7–21.1]; P<.0001). Adjustments for age, sex, dose, and prior vaccination did not have a significant impact on this relationship.

There were 2 non-vaccine-associated serious adverse events reported. One occurred in a 33-year-old woman, who developed a Staphylococcus aureus leg infection after a surgical procedure 77 days after vaccination, and the other occurred in a 42-year-old woman, who attempted suicide 67 days after vaccination.

Results of immunogenicity testing by dose group are shown in table 1. As expected, the antibody levels declined substantially before receipt of the third dose (∼6 months after the second dose). A significant proportion of participants, however, retained MN and HAI GMTs ⩾1:40 at the time of the third dose (7%–14% in the 7.5- and 15-µg-dose groups and 21%–34% in the 45- and 90-µg-dose groups).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1.

Reverse cumulative distribution curves for serum samples collected 28 days and 6 months after vaccination. The figures show the proportion of participants achieving or exceeding the titer indicated on the X-axis, by dose group. A, Results of microneutralization testing 28 days after vaccination. B, Results of microneutralization testing 6 months after vaccination. C, Results of hemagglutination-inhibition testing 28 days after vaccination. D, Results of hemagglutination-inhibition testing 6 months after vaccination. Dose groups are indicated in micrograms.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1.

Geometric mean titers (GMTs) of antibody against the A/Vietnam/1203/04(H5N1) virus and percentage of subjects exhibiting serum antibody GMTs ⩾1:40, as assessed by hemagglutination-inhibition (HAI) and microneutralization (MN) assays.

After the third dose, there was a dose-dependent immune response; MN GMTs were generally higher than HAI GMTs, for each dose group. As shown in table 1, GMTs 28 days after the third dose were highest for the 90- and 45-µg-dose groups. The ratio of post-third-doseMNGMTsto those seen after the second dose were statistically significant for all vaccine concentrations (1.8 -2.3-fold; P<.001). By HAI, the fold-change in GMTs was less impressive (1.1-1.3-fold increase). Six months after the third dose, MN GMTs were still significantly greater than those seen before the third dose, for all dose levels (1.2-1.5-fold increase; P<.01).

Figure 1 presents the reverse cumulative distribution curves for MN and HAI GMTs seen 28 days and 6 months after vaccination, by dose group. MN titers of ⩾1:20 were seen in 92% of recipients of the 90-µg dose, 86% of recipients of the 45-µg dose, 61% of recipients of the 15-µg dose, and 58% of recipients of the 7.5-µg dose. Over all of the vaccine doses, the proportion of participants with GMTs ⩾1:40 was greater by MN than by HAI 28 days and 6 months after the third dose. Overall, the MN and HAI antibody data correlated well (Spearman correlation coefficient of 0.582).

Both univariate and multivariate comparisons (adjusted for age, sex, receipt of prior vaccination, and dose) indicated that participants who seroconverted after 2 doses of vaccine (MN GMT of ⩾1:40) were more likely to do so after the third dose (P<.001); dose concentration did not alter this association. Among 234 of 333 participants who did not respond to the second dose, 88 (38%) responded to the third dose (MN GMT of ⩾1:40). Overall, younger (as enrolled in a stratified manner a priori into those 18–39 and those 40–64 years of age) female participants developed ∼26% higher MN GMTs than did male participants and older female participants (P = .05).

Discussion. In this report, we present the first information regarding the safety and immunogenicity of 3 successive doses of an H5N1-containing influenza vaccine, the strain currently considered to be the most likely to result in a human pandemic, should it occur. A third dose of this unadjuvanted vaccine stimulates an MN response that was substantial and of greater magnitude than that seen 28 days after the second dose of the primary series, in the same population [3]. This suggests that immune memory was retained after 2 doses and that an anamnestic response occurred. In addition, we provide evidence of somewhat greater persistence of antibodies 6 months after the third dose compared with 6 months after a 2-dose regimen. These data suggest that delayed use of a third dose of this vaccine may be useful in pandemic planning, perhaps even as a component of the annual influenza vaccine.

Only one other study has evaluated the immune responses to 3 doses of H5 in humans. That study included 15 adults and evaluated the heterologous anti-H5N1 immune response after 2 doses of an H5N3 vaccine adjuvanted with MF59 (an oil-inwater emulsion) followed by a third vaccination 16 months later. Immune boosting was evident in 8 (53%) of 15 and 100% of participants by HAI and MN assays, respectively, but responses in among 11 adults who received unadjuvanted H5N3 vaccine was much poorer [5]. This information and the data presented here suggest that delayed boosting may be a useful strategy to combat the emergence of new or unexpected pandemic strains.

Our observation that a third vaccine dose significantly enhances the antibody response suggests a potential strategy to overcome the poor immunogenicity of current H5-containing vaccines [11]. In addition, we did not detect any unexpected severity or pattern of adverse events with 3 inoculations regardless of dose, thus providing some basis for further work to evaluate delayed-boosting strategies for putative pandemic influenza vaccines. Such studies might include evaluation of homologous and heterologous immunogenicity after boosting, different priming schedules with varying boost intervals thereafter, and combinations of vaccine strains (e.g., H5N3, H5N3, and then H5N1). It is possible that a combination of approaches will be required to maximize the immune response generated by newer pandemic vaccines.

• Received June 21, 2007.
• Accepted August 7, 2007.

• © 2008 by the Infectious Diseases Society of America