Professional Documents
Culture Documents
H5N1 avian influenza viruses continue to cause disease and an increasing number of cases have been noted in
in humans. Although cases have been reported in 12 Europe and Africa, elevating concern that the virus will
countries, most of the fatalities have been reported from continue to spread to human populations, acquire the
Vietnam, Indonesia, Egypt, Thailand, China, and Tur- ability to spread from person to person, and cause a pan-
key [1]. Outbreaks in birds have been reported in Asia, demic. Concern regarding this spread has been height-
ened by the high mortality rates associated with infec-
Received 1 August 2007; accepted 27 September 2007; electronically published
tion in humans [2, 3]. Vaccination remains the most
8 February 2008. attractive strategy for preventing or limiting the spread
Potential conflicts of interest: D. I. B. receives funding for clinical studies from
Medimmune, GlaxoSmith-Kline, and Sanofi. K.M.E. receives research funding from
of avian influenza in human populations.
Sanofi and Medimmune, and is a consultant for PATH. C.L.D. worked for Chiron Early trials with H5N1 vaccine and other avian influ-
Vaccines until 1997, but currently has no relationship or holdings with them. R.B.
is a consultant for Medimmune and Chiron; a speaker for Merck, Medimmune and
enza vaccines suggested that at least 2 doses of vaccine
Sanofi; and receives funding from Medimmune and Merck. Other authors report no were needed to induce acceptable antibody responses
relevant conflicts.
Financial support: National Institutes of Health (N01 AI 25459, to D.I.B.; AI
and that the antigen content would need to be increased
25462, to K.M.E.; and AI 25464, to R.B.). beyond the amount routinely used in seasonal influenza
Reprints or correspondence: Dr. David Bernstein, 3333 Burnet Ave, ML 6014,
Cincinnati, OH 45229 (david.bernstein@cchmc.org).
vaccines, to increase the magnitude and frequency of the
The Journal of Infectious Diseases 2008; 197:667–75 responses [4 – 6]. In the most recently published study,
© 2008 by the Infectious Diseases Society of America. All rights reserved. Treanor et. al. [5] administered 2 doses of a subvirion
0022-1899/2008/19705-0009$15.00
DOI: 10.1086/527489 H5N1 vaccine at 7.5, 15, 45, or 90 g per dose to healthy
adults. All doses were well tolerated but doses of less than 90 g METHODS
induced serum hemagglutination-inhibition (HAI) titers or
neutralization titers of 肁40 in less than 50% of subjects. Among Subjects. Healthy, nonpregnant adults (age 18 – 64 years) were
those who received the highest dose of vaccine, an HAI titer of eligible for participation. Subjects were excluded if they had a
肁40 was achieved in 58% of subjects, whereas a neutralization known allergy to eggs or other vaccine components; had previ-
titer of 肁40 was observed in 54%. ously received any influenza A/H5 vaccine; were immunosup-
Other reports have indicated that adjuvants enhance the pressed; were known to have HIV, hepatitis B, or hepatitis C
immune response to subvirion H5N1 vaccines [7, 4, 8]. infection; had a history of Guillain-Barré syndrome; had re-
Aluminum-containing adjuvants, the only adjuvants approved ceived blood products in the past 3 months; had received killed
for human use in the United States, have been used routinely in vaccines in the past 2 weeks or live vaccines in the past 4 weeks;
a number of vaccines and have been investigated in conjunction had an acute or chronic illness that would interfere with evalu-
with inactivated influenza vaccines. For seasonal influenza vac- ations or increase risk; or had a history of severe reactions after
cines, aluminum hydroxide has not been shown to clearly in- influenza vaccination. Subjects were also excluded if they had
crease immunogenicity. In one study involving an inactivated laboratory values outside of the normal range on their screening
H5N1 influenza vaccine, the addition of aluminum hydroxide blood tests, including hemoglobin level, white blood cell count,
modestly increased the HAI and neutralization titers (not a sta- platelet count, alanine aminotransferase level, and creatinine
tistically significant increase) at the highest vaccine dose tested level. Eligible subjects were enrolled between March 6 and
(30 g), but not at the lower doses tested (7.5 and 15 g) [4]. March 27, 2006. The protocol and consent forms were approved
Other adjuvants, including MF59, significantly increase the by the institutional review board of each participating center.
antibody titers to A/H5N3 [8], and A/H9N2 [7]. MF59 is an Vaccine and adjuvants. Chiron Vaccines (now part of No-
oil-in-water emulsion adjuvant that has been approved for use vartis) prepared the monovalent inactivated subvirion vaccine.
with seasonal influenza vaccines in several countries in Europe The virus used to prepare the working seed and the vaccine was
since 1997. MF59 has been shown to improve antibody re- produced by reverse genetics using the modified hemagglutinin
sponses to influenza, herpes simplex, cytomegalovirus, and HIV and unaltered neuraminidase encoding genes from the influenza
vaccines [9]. In the most recent avian influenza vaccine study A/Vietnam/1203/2004(H5N1) strain and all other genes from
involving MF59 [7], HAI and neutralization titers to A/H9N2 A/PuertoRico/8/34 [10]. The vaccine virus was then grown in
were higher after the first and second doses in the groups that embryonated eggs by use of a process similar to that used for
received vaccine with MF59, compared with the group that re- seasonal vaccine preparation. The vaccine was formulated to
ceived vaccine alone, following doses of 3.75, 7.5, 15, or 30 g. In contain a hemagglutinin concentration of 60 g/mL.
this study, there was no evidence of a dose effect. Administration of the vaccine at different doses was accom-
In the current study, we evaluated a subvirion influenza plished by varying the volume of the vaccine and adjuvant mix-
A/H5N1 vaccine at doses from 7.5 g to 45 g with and without tures. Vaccine and adjuvant were prepared daily in the investi-
aluminum hydroxide or MF59. Our results with respect to the gational pharmacy by withdrawing 0.7 or 0.8 mL of antigen from
safety and immunogenicity of these vaccines are reported here. 3-mL vials containing 60g/mL of antigen and adding this to
vials with equal volumes of adjuvant (0.7 or 0.8 mL) or using the vaccine, which was available for the trial as a single formula-
antigen alone. After mixing, the product was administered tion of 60 g/mL, and the expected tolerable volumes and con-
within 8 h, and it was administered within 10 min of drawing the centrations of MF59 and aluminum hydroxide. The sample size
mixture into the syringe. Vaccine and adjuvant were adminis- was weighted to favor doses that were deemed to be the most
tered at the dose and volume shown in table 1. It should be noted likely to be implemented in the future, to obtain more precise
that decreases in the volume administered for doses with re- estimates of the responses in those dose groups. The study was
duced antigen content led to decreasing amounts of MF59 and not designed to test a specific hypothesis. However, sample sizes
aluminum hydroxide being administered with decreasing doses of 90 and 60 subjects in the groups of greatest interest provided
of antigen, as described above. The MF59 content at the higher 80% power to detect a difference in the proportion of responders
dose was 9.75 mg squalene, 1.68 mg polysorbate 80, 1.68 mg of 50% and 51%, respectively, when compared with a group of
sorbitan trioleate, 0.96 mg sodium citrate, and 0.06 mg cit- 30, assuming a 20% response rate for the smaller group. Addi-
ric acid; the lower dose contained half these amounts. The alu-
tionally, sample sizes of 90 and 60 provided at least 99% and 95%
minum hydroxide content ranged from 0.35 mg to 0.88 mg
probabilities, respectively, of observing a vaccine-related event,
(table 1).
assuming a true occurrence rate of at least 5.0%. In addition,
Study design. This multicenter randomized, double-blind,
these sample sizes provided the full widths of, at most, 21.5%
placebo-controlled trial was conducted at 4 sites in the United
and 26.4% respectively, based on a 95% Blyth-Still confidence
States: Cincinnati Children’s Hospital Medical Center, Cincin-
interval of an observed event rate.
nati, Ohio; Vanderbilt University Medical Center, Nashville,
After randomization, an unblinded member of the study team
Tennessee; Stanford University School of Medicine, Stanford,
who did not otherwise participate in the trial administered the
California; and St. Louis University, St. Louis, Missouri. After
providing informed consent, eligible subjects were randomly as- injection. Subjects were observed for 15 min; they then com-
signed to 1 of 9 groups: saline placebo (n ⫽ 30); influenza vac- pleted a memory aid to record local and systemic adverse events,
cine alone at 45, 30, or 15 g per dose (n ⫽ 90, 60, and 30, as well as their daily temperatures for 7 days. Subjects were called
respectively); influenza vaccine at 15 or 7.5 g per dose, com- 1–3 days and 14 –21 days after each vaccination to elicit infor-
bined with MF59 (n ⫽ 30, for both groups); or influenza vac- mation about any adverse events. Subjects also returned to the
cine at 30, 15, or 7.5 g per dose, combined with aluminum clinic 7–9 days after each vaccination for assessment and to re-
hydroxide (n ⫽ 60, 30, and 30, respectively) (table 1). Each sub- peat the same laboratory tests outlined earlier. Approximately 28
ject received 2 intramuscular injections of an identical formula- days after each vaccination, and again at 6 months after the sec-
tion in the deltoid area, administered approximately 28 days ond vaccination, serum samples were obtained for antibody as-
apart. says. All 7-day safety data available after the subject had received
The choice of doses and the combinations of vaccine with the first dose of vaccine were reviewed prior to administering the
adjuvant preparations used in the study were largely driven by second dose of vaccine.
NOTE. Data are no. (%) of subjects who had a reaction or abnormal laboratory value after either the first or second injection, unless otherwise indicated. ALT,
alanine aminotransferase; WBC, white blood cell.
a
P values were calculated with the 2 test, except for fever, which was calculated by the Fisher exact test.
b
Most severe score after either the first or second vaccination.
c
The P value for the mean pain score was calculated by analysis of variance.
Antibody assays. Microneutralization (MN) and hemagglutin- als being randomized, only the total number that investigators
ation-inhibition (HAI) assays were performed at Southern Re- planned to randomize. The site staff who evaluated vaccine re-
search Institute using genetically modified reassortant, rgA/Viet- actogenicity were fully blinded to the dose and type of vaccine
nam/1203/2004x A/PR/8/34, as described elsewhere [11, 12]. MN given.
and HAI assays with horse erythrocytes were performed as de- The proportions of solicited adverse events (mild, moderate,
scribed elsewhere [5], except that the same starting dilution was or severe) after either the first or second dose of vaccine were
defined as 1:10 rather than 1:20, and samples that were negative based on the most severe response reported. The proportion of
were assigned a titer of 5. Development of a titer of 肁40 in both the individuals with an elevated oral temperature was compared
MN and HAI assays was predefined as indicative of a “responder.” across 9 treatment groups by use of the Fisher exact test, and the
Detectable responses were defined as a titer 肁10. For both assays, proportions of the remaining local and systemic reactions were
serum samples were tested in duplicate in independent assays, and compared by use of the 2 test. Pain at the injection site was also
samples that gave more than 2-fold different results in replicate analyzed by assigning a numerical scale (0, 1, 2, or 3) to the level
samples were retested. of response (none, mild, moderate, or severe). The relationship
Statistical analysis. Data collection for this study was con- of pain to dose levels and adjuvant effects was explored with the
ducted using the Statistical and Data Coordinating Center’s general linear model, and the result was also verified by use of
(SDCC) Internet-based Electronic Data Capture system. The only an ordinal scale and testing with cumulative logistic regression.
SDCC was also responsible for the randomization. Because of Analytical results for geometric mean titers (GMTs) were de-
the limited supply of study vaccine, randomization tables were termined by converting to a logarithmic scale and assuming nor-
established at the SDCC each night for the following day, on the mality or using rank order methods. Exact confidence intervals
basis of only the cumulative enrollment for all sites and the ex- were reported for all proportional endpoints. For subjects with
pected number of new enrollees at each site. At the time of ran- and subjects without preexisting antibodies to the H5N1 virus,
domization, the SDCC had no information about the individu- geometric mean titers were compared using the Wilcoxon rank
Subjects with
Antibody Subjects titer 肁40, no. GMT
response, dose Adjuvant tested, no. (% [95% CI]) (95% CI)
Hemagglutination
inhibiting antibody
0 None 28 0 (0 [0–12]) 5.8 (4.9–6.9)
7.5 Aluminum hydroxide 29 1 (3 [0–18]) 6.1 (4.7–7.9)
7.5 MF59 30 7 (23 [10–42]) 13.3 (8.2–21.7)
15 None 29 7 (24 [10–44]) 10.2 (6.3–16.6)
15 Aluminum hydroxide 28 2 (7 [1–24]) 7.1 (5.4–9.3)
15 MF59 32 20 (63 [44–70]) 32.6 (20–53)
30 none 60 11 (18 [10–30]) 10.7 (7.9–14.3)
30 Aluminum hydroxide 58 8 (14 [6–25]) 9.1 (6.8–12.2)
45 None 84 24 (29 [19–39]) 15.7 (11.6–21.2)
Microneutralization
0 None 28 1 (4 [0–18]) 5.9 (4.9–7.2)
7.5 Aluminum hydroxide 29 4 (14 [4–32]) 10.8 (7.6,15.3)
7.5 MF59 30 11 (37 [20–56]) 25.7 (17.4–37.8)
15 None 29 8 (28 [13–47]) 15.8 (10.7–23.3)
15 Aluminum hydroxide 28 5 (18 [6–37]) 12.5 (8.6–18.2)
15 MF59 32 26 (81 [64–93]) 63.0 (45.1–88.1)
30 None 60 17 (28 [17–41]) 18.0 (13.8–23.6)
30 Aluminum hydroxide 58 14 (24 [14–37]) 16.2 (12.4–21.2)
45 None 84 40 (48 [37–59]) 29.2 (22.8–37.4)
test, and the proportions of 4-fold increases from the baseline the groups that received aluminum hydroxide, compared with
were compared using the Fisher exact test. All analyses were con- the groups that received no adjuvant, and the incidence and se-
ducted using SAS (version 8.2; SAS Institute). verity of pain increased further in the groups that received
MF59. Therefore, pain was most common after a dose of 15 g
RESULTS of vaccine with MF59. The pain was, however, mild to moderate
and lasted for 1–2 days for the majority of subjects. The inci-
A total of 394 subjects received at least 1 dose of vaccine and 382 dence of tenderness was also similarly related to vaccine dose
received 2 doses (figure 1). The reasons that subjects were with- level and adjuvant. Neither local or systemic adverse events in-
drawn from the study and their group assignment is also shown creased after the second dose, compared with the first. No sub-
in figure 1. The demographic characteristics of the population ject withdrew from the study because of the intensity of local
were comparable among the vaccine groups, except for a signif- reactions.
icant difference noted in the sex distribution (P ⬍ .01 by 2 Immunogenicity. Antibody responses were measured by
[table 2]). HAI and MN at baseline, at 28 days after each dose, and at 6
Safety. There were no serious adverse events associated with months after the second dose. After the first dose of vaccine,
the vaccine. As shown in table 3, local reactions, especially pain 7%–29% of subjects in each group had detectable HAI antibody
and tenderness were common, whereas systemic adverse events (i.e., titer 肁10), and 18%– 66% had detectable MN antibody
occurred in as few as 0%–3% for fever and as many as 13%– 44% (data not shown). However, antibody at the predefined end-
for headache in the 7 days following vaccination. No significant point titer of 肁40 was observed in only 0%–22% of subjects, as
differences in vaccine reactogenicity were seen among the measured by HAI, and in 3%–21% of subjects, as measured by
groups, except for local pain and tenderness. The incidence and MN (data not shown). The highest percentage of responders was
severity of pain was related to both the vaccine dose and the seen in the group receiving 15 g of vaccine with MF59. The
adjuvant. The incidence and severity of pain increased as dose geometric mean titer after the first dose was slightly higher in the
levels increased in each adjuvant group (P ⫽ .01, Type III group that received 45 g of vaccine alone (titer 10.8), compared
F-test). The incidence and severity of pain were also increased in with the group that received 15 g of vaccine with MF59 (titer 9.5).
After the second dose of vaccine, antibody levels increased antibody at a titer 肁40 in a range from 24% at the lowest dose
in all groups. HAI antibody was detected (titer 肁10) in 10%– (15 g) to 29% at the highest dose (45 g). The percentages of
72% of subjects and MN antibody was detected in 45%–97% subjects achieving MN titer 肁40 and the MN GMTs were
of subjects (data not shown). As shown in table 4, the highest somewhat higher, compared with the HAI responses. Thus,
antibody response was seen in the group that received 15 g the percentage of subjects with MN titer 肁40 in the group
of vaccine with MF59, in which 63% of subjects achieved an receiving 15 g of vaccine with MF59 was 81%, with a GMT
HAI titer 肁40 (GMT 32.6). The vaccine alone induced HAI of 63.0. The addition of aluminum hydroxide did not increase
Table 5. Comparison of geometric mean titer (GMT) between those with baseline titer <10
and those with baseline titer 肁10.