MAJOR ARTICLE

Effects of Adjuvants on the Safety and

Immunogenicity of an Avian Influenza
H5N1 Vaccine in Adults
David I. Bernstein,1 Kathryn M. Edwards,2 Cornelia L. Dekker,3 Robert Belshe,4 Helen K. B. Talbot,5 Irene L. Graham,4
Diana L. Noah,6 Fenhua He,7 and Heather Hill7
1
Division of Infectious Diseases, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio; 2Pediatric Infectious Diseases and 5Medicine,
Vanderbilt University Medical Center, Nashville, Tennessee; 3Pediatric Infectious Diseases, Stanford University School of Medicine, Stanford,
California; 4Infectious Diseases and Immunology, St. Louis University, St. Louis, Missouri; 6Southern Research Institute, Birmingham, Alabama;
7
EMMES, Rockville, Maryland.

Background. Influenza A H5N1 viruses pose a significant threat to human health.

Methods. We conducted a multicenter, randomized, double-blind study in 394 healthy adults. Subjects were
randomly assigned to receive 2 intramuscular doses of either saline placebo; influenza A/Vietnam/1203/2004(H5N1)
vaccine alone at 45, 30, or 15 ␮g per dose; vaccine at 15 or 7.5 ␮g per dose with MF59; or vaccine at 30, 15, or 7.5 ␮g
per dose with aluminum hydroxide. Subjects were followed up for safety and blood samples were obtained to deter-
mine antibody responses.
Results. The vaccine formulations were well tolerated but local adverse effects were common; the incidence of
these effects increased in a dose-dependent manner and was increased by the addition of adjuvants. The addition of
MF59 increased the antibody response, whereas the addition of aluminum hydroxide did not. The highest antibody
responses were seen in the group that received 15 ␮g of vaccine per dose with MF59, in which 63% of subjects achieved
the predetermined endpoint (hemagglutination-inhibition titer 肁40) 28 days after the second dose, compared with
29% in the group that received the highest dose (45 ␮g per dose) of vaccine alone.
Conclusions. A 2-dose regimen of subvirion influenza A (H5N1) vaccine was well tolerated. The antibody re-
sponses to 15 ␮g of A/H5 vaccine with MF59 were higher than the responses to 45 ␮g of vaccine alone.
Trial registration. ClincalTrials.gov identifier: http://www.clinicaltrials.gov/ct2/show/NCT00280033?term⫽
NCT00280033&rank⫽1NCT00280033.

H5N1 avian influenza viruses continue to cause disease
in humans. Although cases have been reported in 12
countries, most of the fatalities have been reported from
Vietnam, Indonesia, Egypt, Thailand, China, and Tur-
key [1]. Outbreaks in birds have been reported in Asia,
Received 1 August 2007; accepted 27 September 2007; electronically published
8 February 2008.
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
Sanofi and Medimmune, and is a consultant for PATH. C.L.D. worked for Chiron
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
Sanofi; and receives funding from Medimmune and Merck. Other authors report no
relevant conflicts.
Financial support: National Institutes of Health (N01 AI 25459, to D.I.B.; AI
25462, to K.M.E.; and AI 25464, to R.B.).
Reprints or correspondence: Dr. David Bernstein, 3333 Burnet Ave, ML 6014,
Cincinnati, OH 45229 (david.bernstein@cchmc.org).
The Journal of Infectious Diseases 2008; 197:667–75
© 2008 by the Infectious Diseases Society of America. All rights reserved.
0022-1899/2008/19705-0009$15.00
DOI: 10.1086/527489
and an increasing number of cases have been noted in
Europe and Africa, elevating concern that the virus will
continue to spread to human populations, acquire the
ability to spread from person to person, and cause a pan-
demic. Concern regarding this spread has been height-
ened by the high mortality rates associated with infec-
tion in humans [2, 3]. Vaccination remains the most
attractive strategy for preventing or limiting the spread
of avian influenza in human populations.
Early trials with H5N1 vaccine and other avian influ-
enza vaccines suggested that at least 2 doses of vaccine
were needed to induce acceptable antibody responses
and that the antigen content would need to be increased
beyond the amount routinely used in seasonal influenza
vaccines, to increase the magnitude and frequency of the
responses [4 – 6]. In the most recently published study,
Treanor et. al. [5] administered 2 doses of a subvirion
H5N1 vaccine at 7.5, 15, 45, or 90 ␮g per dose to healthy

Evaluation of Influenza H5N1 Vaccine ● JID 2008:197 (1 March) ● 667

 Table 1. Dose and volume of vaccine and adjuvant per study design.

Hemagglutinin Al3⫹ or MF59

Subjects, content per content per dose, Volume of
Study group, dose no. dose, ␮g mg or mL injection, mL
Unadjuvanted vaccine
45 ␮g 90 45 ... 0.75
30 ␮g 60 30 ... 0.5
15 ␮g 30 15 ... 0.25
Vaccine with MF59
15 ␮g 30 15 0.25 0.5
7.5 ␮g 30 7.5 0.125 0.25
Vaccine with aluminum
hydroxide
30 ␮g 60 30 0.350 1.0
15 ␮g 30 15 0.175 0.5
7.5 ␮g 30 7.5 0.0875 0.25
Placebo (saline) 30 ... ... 0.5

adults. All doses were well tolerated but doses of less than 90 ␮g
induced serum hemagglutination-inhibition (HAI) titers or
neutralization titers of 肁40 in less than 50% of subjects. Among
those who received the highest dose of vaccine, an HAI titer of
肁40 was achieved in 58% of subjects, whereas a neutralization
titer of 肁40 was observed in 54%.
Other reports have indicated that adjuvants enhance the
immune response to subvirion H5N1 vaccines [7, 4, 8].
Aluminum-containing adjuvants, the only adjuvants approved
for human use in the United States, have been used routinely in
a number of vaccines and have been investigated in conjunction
with inactivated influenza vaccines. For seasonal influenza vac-
cines, aluminum hydroxide has not been shown to clearly in-
crease immunogenicity. In one study involving an inactivated
H5N1 influenza vaccine, the addition of aluminum hydroxide
modestly increased the HAI and neutralization titers (not a sta-
tistically significant increase) at the highest vaccine dose tested
(30 ␮g), but not at the lower doses tested (7.5 and 15 ␮g) [4].
Other adjuvants, including MF59, significantly increase the
antibody titers to A/H5N3 [8], and A/H9N2 [7]. MF59 is an
oil-in-water emulsion adjuvant that has been approved for use
with seasonal influenza vaccines in several countries in Europe
since 1997. MF59 has been shown to improve antibody re-
sponses to influenza, herpes simplex, cytomegalovirus, and HIV
vaccines [9]. In the most recent avian influenza vaccine study
involving MF59 [7], HAI and neutralization titers to A/H9N2
were higher after the first and second doses in the groups that
received vaccine with MF59, compared with the group that re-
ceived vaccine alone, following doses of 3.75, 7.5, 15, or 30 ␮g. In
this study, there was no evidence of a dose effect.
In the current study, we evaluated a subvirion influenza
A/H5N1 vaccine at doses from 7.5 ␮g to 45 ␮g with and without
aluminum hydroxide or MF59. Our results with respect to the
safety and immunogenicity of these vaccines are reported here.
METHODS
Subjects. Healthy, nonpregnant adults (age 18 – 64 years) were
eligible for participation. Subjects were excluded if they had a
known allergy to eggs or other vaccine components; had previ-
ously received any influenza A/H5 vaccine; were immunosup-
pressed; were known to have HIV, hepatitis B, or hepatitis C
infection; had a history of Guillain-Barré syndrome; had re-
ceived blood products in the past 3 months; had received killed
vaccines in the past 2 weeks or live vaccines in the past 4 weeks;
had an acute or chronic illness that would interfere with evalu-
ations or increase risk; or had a history of severe reactions after
influenza vaccination. Subjects were also excluded if they had
laboratory values outside of the normal range on their screening
blood tests, including hemoglobin level, white blood cell count,
platelet count, alanine aminotransferase level, and creatinine
level. Eligible subjects were enrolled between March 6 and
March 27, 2006. The protocol and consent forms were approved
by the institutional review board of each participating center.
Vaccine and adjuvants. Chiron Vaccines (now part of No-
vartis) prepared the monovalent inactivated subvirion vaccine.
The virus used to prepare the working seed and the vaccine was
produced by reverse genetics using the modified hemagglutinin
and unaltered neuraminidase encoding genes from the influenza
A/Vietnam/1203/2004(H5N1) strain and all other genes from
A/PuertoRico/8/34 [10]. The vaccine virus was then grown in
embryonated eggs by use of a process similar to that used for
seasonal vaccine preparation. The vaccine was formulated to
contain a hemagglutinin concentration of 60 ␮g/mL.
Administration of the vaccine at different doses was accom-
plished by varying the volume of the vaccine and adjuvant mix-
tures. Vaccine and adjuvant were prepared daily in the investi-
gational pharmacy by withdrawing 0.7 or 0.8 mL of antigen from
3-mL vials containing 60␮g/mL of antigen and adding this to

668 ● JID 2008:197 (1 March) ● Bernstein et al.

 Figure 1. Trial profile and reasons for subject withdrawal
vials with equal volumes of adjuvant (0.7 or 0.8 mL) or using
antigen alone. After mixing, the product was administered
within 8 h, and it was administered within 10 min of drawing the
mixture into the syringe. Vaccine and adjuvant were adminis-
tered at the dose and volume shown in table 1. It should be noted
that decreases in the volume administered for doses with re-
duced antigen content led to decreasing amounts of MF59 and
aluminum hydroxide being administered with decreasing doses
of antigen, as described above. The MF59 content at the higher
dose was 9.75 mg squalene, 1.68 mg polysorbate 80, 1.68 mg
sorbitan trioleate, 0.96 mg sodium citrate, and 0.06 mg cit-
ric acid; the lower dose contained half these amounts. The alu-
minum hydroxide content ranged from 0.35 mg to 0.88 mg
(table 1).
Study design. This multicenter randomized, double-blind,
placebo-controlled trial was conducted at 4 sites in the United
States: Cincinnati Children’s Hospital Medical Center, Cincin-
nati, Ohio; Vanderbilt University Medical Center, Nashville,
Tennessee; Stanford University School of Medicine, Stanford,
California; and St. Louis University, St. Louis, Missouri. After
providing informed consent, eligible subjects were randomly as-
signed to 1 of 9 groups: saline placebo (n ⫽ 30); influenza vac-
cine alone at 45, 30, or 15 ␮g per dose (n ⫽ 90, 60, and 30,
respectively); influenza vaccine at 15 or 7.5 ␮g per dose, com-
bined with MF59 (n ⫽ 30, for both groups); or influenza vac-
cine at 30, 15, or 7.5 ␮g per dose, combined with aluminum
hydroxide (n ⫽ 60, 30, and 30, respectively) (table 1). Each sub-
ject received 2 intramuscular injections of an identical formula-
tion in the deltoid area, administered approximately 28 days
apart.
The choice of doses and the combinations of vaccine with
adjuvant preparations used in the study were largely driven by
the vaccine, which was available for the trial as a single formula-
tion of 60 ␮g/mL, and the expected tolerable volumes and con-
centrations of MF59 and aluminum hydroxide. The sample size
was weighted to favor doses that were deemed to be the most
likely to be implemented in the future, to obtain more precise
estimates of the responses in those dose groups. The study was
not designed to test a specific hypothesis. However, sample sizes
of 90 and 60 subjects in the groups of greatest interest provided
80% power to detect a difference in the proportion of responders
of 50% and 51%, respectively, when compared with a group of
30, assuming a 20% response rate for the smaller group. Addi-
tionally, sample sizes of 90 and 60 provided at least 99% and 95%
probabilities, respectively, of observing a vaccine-related event,
assuming a true occurrence rate of at least 5.0%. In addition,
these sample sizes provided the full widths of, at most, 21.5%
and 26.4% respectively, based on a 95% Blyth-Still confidence
interval of an observed event rate.
After randomization, an unblinded member of the study team
who did not otherwise participate in the trial administered the
injection. Subjects were observed for 15 min; they then com-
pleted a memory aid to record local and systemic adverse events,
as well as their daily temperatures for 7 days. Subjects were called
1–3 days and 14 –21 days after each vaccination to elicit infor-
mation about any adverse events. Subjects also returned to the
clinic 7–9 days after each vaccination for assessment and to re-
peat the same laboratory tests outlined earlier. Approximately 28
days after each vaccination, and again at 6 months after the sec-
ond vaccination, serum samples were obtained for antibody as-
says. All 7-day safety data available after the subject had received
the first dose of vaccine were reviewed prior to administering the
second dose of vaccine.
Evaluation of Influenza H5N1 Vaccine ● JID 2008:197 (1 March) ● 669
Table 2. Demographic characteristics of study subjects, according to vaccine group.

Vaccine alone Vaccine with aluminum hydroxide Vaccine with MF59

All Placebo 15 ␮g 30 ␮g 45 ␮g 7.5 ␮g 15 ␮g 30 ␮g 7.5 ␮g 15 ␮g

Characteristic (N ⫽ 394) (n ⫽ 29) (n ⫽ 30) (n ⫽ 61) (n ⫽ 90) (n ⫽ 30) (n ⫽ 31) (n ⫽ 59) (n ⫽ 32) (n ⫽ 32)
Sex
Male 151 (38) 14 (48) 8 (27) 15 (25) 39 (43) 13 (43) 8 (26) 18 (31) 17 (53) 19 (59)
Female 243 (62) 15 (52) 22 (73) 46 (75) 51 (57) 17 (57) 23 (74) 41 (69) 15 (47) 13 (41)
Race
Asian 14 (4) 0 (0) 1 (3) 4 (7) 3 (3) 2 (7) 0 (0) 3 (5) 0 (0) 1 (3)
African American 20 (5) 0 (0) 4 (13) 2 (3) 5 (6) 1 (3) 2 (6) 3 (5) 1 (3) 2 (6)
White 350 (89) 27 (93) 23 (77) 53 (87) 80 (89) 26 (87) 29 (94) 53 (90) 31 (97) 28 (88)
Other, unknown, or multiracial 10 (3) 2 (7) 2 (7) 2 (3) 2 (2) 1 (3) 0 (0) 0 (0) 0 (0) 1 (3)
Age, years
Mean ⫾ SD 41.7 ⫾ 12.2 42.8 ⫾ 12.7 39.8 ⫾ 13.6 39.8⫾ 12.4 42.5⫾ 12.5 43.3 ⫾ 11.8 43.7 ⫾ 12.7 40.1 ⫾ 12.1 42.4 ⫾ 10.3 43.0 ⫾ 11.6
Median 42.2 47.1 35.1 38.4 43.7 41.5 44.8 41.3 45.6 41.7
Range 18.9–64.9 22.5–64.4 19.8–61.4 18.9–62.9 19.0–64.0 24.3–62.8 19.4–63.8 19.2–62.7 20.0–55.7 21.2–64.9

NOTE. Data are no. (%) of subjects, unless otherwise indicated.

Table 3. Incidence of adverse events after the first or second dose of vaccine, according to vaccine group.

Vaccine with Vaccine

Vaccine alone aluminum hydroxide with MF59

Placebo 15 ␮g 30 ␮g 45 ␮g 7.5 ␮g 15 ␮g 30 ␮g 7.5 ␮g 15 ␮g

Adverse event (n ⫽ 29) (n ⫽ 30) (n ⫽ 61) (n ⫽ 90) (n ⫽ 30) (n ⫽ 31) (n ⫽ 59) (n ⫽ 32) (n ⫽ 32) P
Reaction
Local
Pain 6 (21) 6 (20) 15 (25) 34 (38) 10 (33) 12 (39) 30 (51) 18 (56) 22 (69) ⬍.001a
Mean pain scoreb 0.2 0.2 0.3 0.4 0.3 0.4 0.6 0..6 0.8 ⬍.001c
Redness 8 (28) 7 (23) 21 (34) 19 (21) 5 (17) 6 (19) 20 (34) 5 (16) 6 (19) .26
Swelling 6 (21) 5 (17) 12 (20) 16 (18) 3 (10) 4 (13) 13 (22) 5 (16) 4 (13) .90
Tenderness 12 (41) 13 (43) 30 (49) 47 (52) 16 (53) 18 (58) 46 (78) 23 (72) 22 (69) .003
Systemic
Fever (肁37.8°C) 1 (3) 0 (0) 1 (2) 2 (2) 0 (0) 0 (0) 0 (0) 0 (0) 1 (3) .78a
Feverishness 2 (7) 1 (3) 3 (5) 8 (9) 2 (7) 1 (3) 1 (2) 4 (13) 3 (9) .55
Malaise 7 (24) 6 (20) 19 (31) 22 (24) 8 (27) 7 (23) 16 (27) 8 (25) 8 (25) .99
Myalgia 4 (14) 3 (10) 12 (20) 12 (13) 8 (27) 5 (16) 14 (24) 9 (28) 10 (31) .22
Headache 9 (31) 4 (13) 22 (36) 26 (29) 11 (37) 11 (35) 26 (44) 12 (38) 9 (28) .24
Nausea 4 (14) 4 (13) 8 (13) 11 (12) 3 (10) 1 (3) 9 (15) 4 (13) 4 (13) .93
Abnormal laboratory values
Hemoglobin level 0 (0) 0 (0) 1 (2) 5 (6) 0 (0) 0 (0) 3 (5) 0 (0) 0 (0) .501
WBC count 1 (3) 4 (13) 3 (5) 5 (6) 0 (0) 2 (6) 2 (3) 2 (6) 1 (3) .633
Creatinine level 2 (7) 1 (3) 1 (2) 1 (1) 1 (3) 2 (6) 2 (3) 0 (0) 0 (0) .405
Platelet count 0 (0) 0 (0) 1 (2) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) .772
ALT level 1 (3) 0 (0) 2 (3) 2 (2) 2 (7) 2 (6) 2 (3) 1 (3) 3 (9) .602

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-
ation-inhibition (HAI) assays were performed at Southern Re-
search Institute using genetically modified reassortant, rgA/Viet-
nam/1203/2004x A/PR/8/34, as described elsewhere [11, 12]. MN
and HAI assays with horse erythrocytes were performed as de-
scribed elsewhere [5], except that the same starting dilution was
defined as 1:10 rather than 1:20, and samples that were negative
were assigned a titer of 5. Development of a titer of 肁40 in both the
MN and HAI assays was predefined as indicative of a “responder.”
Detectable responses were defined as a titer 肁10. For both assays,
serum samples were tested in duplicate in independent assays, and
samples that gave more than 2-fold different results in replicate
samples were retested.
Statistical analysis. Data collection for this study was con-
ducted using the Statistical and Data Coordinating Center’s
(SDCC) Internet-based Electronic Data Capture system. The
SDCC was also responsible for the randomization. Because of
the limited supply of study vaccine, randomization tables were
established at the SDCC each night for the following day, on the
basis of only the cumulative enrollment for all sites and the ex-
pected number of new enrollees at each site. At the time of ran-
domization, the SDCC had no information about the individu-
als being randomized, only the total number that investigators
planned to randomize. The site staff who evaluated vaccine re-
actogenicity were fully blinded to the dose and type of vaccine
given.
The proportions of solicited adverse events (mild, moderate,
or severe) after either the first or second dose of vaccine were
based on the most severe response reported. The proportion of
individuals with an elevated oral temperature was compared
across 9 treatment groups by use of the Fisher exact test, and the
proportions of the remaining local and systemic reactions were
compared by use of the ␹2 test. Pain at the injection site was also
analyzed by assigning a numerical scale (0, 1, 2, or 3) to the level
of response (none, mild, moderate, or severe). The relationship
of pain to dose levels and adjuvant effects was explored with the
general linear model, and the result was also verified by use of
only an ordinal scale and testing with cumulative logistic regression.
Analytical results for geometric mean titers (GMTs) were de-
termined by converting to a logarithmic scale and assuming nor-
mality or using rank order methods. Exact confidence intervals
were reported for all proportional endpoints. For subjects with
and subjects without preexisting antibodies to the H5N1 virus,
geometric mean titers were compared using the Wilcoxon rank

Evaluation of Influenza H5N1 Vaccine ● JID 2008:197 (1 March) ● 671

 Table 4. Antibody response in study subjects 28 days after receipt of the second dose of
vaccine.

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)

NOTE. GMT, geometric mean titer.

test, and the proportions of 4-fold increases from the baseline
were compared using the Fisher exact test. All analyses were con-
ducted using SAS (version 8.2; SAS Institute).
RESULTS
A total of 394 subjects received at least 1 dose of vaccine and 382
received 2 doses (figure 1). The reasons that subjects were with-
drawn from the study and their group assignment is also shown
in figure 1. The demographic characteristics of the population
were comparable among the vaccine groups, except for a signif-
icant difference noted in the sex distribution (P ⬍ .01 by ␹2
[table 2]).
Safety. There were no serious adverse events associated with
the vaccine. As shown in table 3, local reactions, especially pain
and tenderness were common, whereas systemic adverse events
occurred in as few as 0%–3% for fever and as many as 13%– 44%
for headache in the 7 days following vaccination. No significant
differences in vaccine reactogenicity were seen among the
groups, except for local pain and tenderness. The incidence and
severity of pain was related to both the vaccine dose and the
adjuvant. The incidence and severity of pain increased as dose
levels increased in each adjuvant group (P ⫽ .01, Type III
F-test). The incidence and severity of pain were also increased in
the groups that received aluminum hydroxide, compared with
the groups that received no adjuvant, and the incidence and se-
verity of pain increased further in the groups that received
MF59. Therefore, pain was most common after a dose of 15 ␮g
of vaccine with MF59. The pain was, however, mild to moderate
and lasted for 1–2 days for the majority of subjects. The inci-
dence of tenderness was also similarly related to vaccine dose
level and adjuvant. Neither local or systemic adverse events in-
creased after the second dose, compared with the first. No sub-
ject withdrew from the study because of the intensity of local
reactions.
Immunogenicity. Antibody responses were measured by
HAI and MN at baseline, at 28 days after each dose, and at 6
months after the second dose. After the first dose of vaccine,
7%–29% of subjects in each group had detectable HAI antibody
(i.e., titer 肁10), and 18%– 66% had detectable MN antibody
(data not shown). However, antibody at the predefined end-
point titer of 肁40 was observed in only 0%–22% of subjects, as
measured by HAI, and in 3%–21% of subjects, as measured by
MN (data not shown). The highest percentage of responders was
seen in the group receiving 15 ␮g of vaccine with MF59. The
geometric mean titer after the first dose was slightly higher in the
group that received 45 ␮g of vaccine alone (titer 10.8), compared
with the group that received 15 ␮g of vaccine with MF59 (titer 9.5).

672 ● JID 2008:197 (1 March) ● Bernstein et al.

Figure 2. Antibody responses over time. The geometric mean antibody titer (GMT) for hemagglutination-inhibiting (HAI) antibody (A) and
microneutralization (MN) antibody (B) is shown for the groups that received the largest vaccine dose without adjuvant, the largest dose with aluminum
hydroxide adjuvant, and the largest dose with MF59 adjuvant. Antibody evaluations were performed prior to vaccination (day 0), 28 days after the first
vaccination (day 28), 28 days after the second vaccination (day 56), and 6 months after the second vaccination (day 208).

After the second dose of vaccine, antibody levels increased
in all groups. HAI antibody was detected (titer 肁10) in 10%–
72% of subjects and MN antibody was detected in 45%–97%
of subjects (data not shown). As shown in table 4, the highest
antibody response was seen in the group that received 15 ␮g
of vaccine with MF59, in which 63% of subjects achieved an
HAI titer 肁40 (GMT 32.6). The vaccine alone induced HAI
antibody at a titer 肁40 in a range from 24% at the lowest dose
(15 ␮g) to 29% at the highest dose (45 ␮g). The percentages of
subjects achieving MN titer 肁40 and the MN GMTs were
somewhat higher, compared with the HAI responses. Thus,
the percentage of subjects with MN titer 肁40 in the group
receiving 15 ␮g of vaccine with MF59 was 81%, with a GMT
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.

Baseline titer ⬍10 Baseline titer 肁10

Test type, time Subjects, no. GMT Subjects, no. GMT Pa Pb

Hemagglutination -inhibiting
antibody
Baseline 367 5.0 11 31.0 ⬍.001 ⬍.001
28 days after dose 1 367 19.3 11 56.3 .02 ⬍.001
Microneutralization
Baseline 335 5.5 43 20.8 ⬍.001 ⬍.001
28 days after dose 1 335 16.8 43 95.0 ⬍.001 ⬍.001
a
By t test.
b
By Wilcoxon rank test.

Evaluation of Influenza H5N1 Vaccine ● JID 2008:197 (1 March) ● 673

 Table 6. Comparison percentage of subjects with a
4-fold increase after 1 dose of vaccine, for subjects with
baseline titer <10 and subjects with baseline titer 肁10.
Subjects with
4-fold increase,
no. (%) of total
subjects
Baseline Baseline
Test type titer ⬍10 titer 肁10
Hemagglutination-inhibiting
antibody 39 (11) 2 (18)
Microneutralization 28 (8) 13 (30)
a
Fisher exact test.
the antibody as measured by HAI or MN to the level observed
after vaccine alone at any dose.
At 6 months after the second dose, antibody titers had de-
creased in all groups. The HAI antibody levels over time are
shown for groups that received the largest vaccine dose without
adjuvant (45␮g), vaccine with aluminum hydroxide (30 ␮g),
and vaccine with MF59 (15 ␮g) (figure 2A). Similar decreases in
MN antibody levels were also seen (figure 2B). At 6 months after
the second dose, HAI responses of 肁40 were observed in 18% of
those who received 45 ␮g of vaccine alone, 5% of those who
received 30 ␮g of vaccine with aluminum hydroxide, and 20% of
those who received 15 ␮g of vaccine with MF59.
At baseline, 0%–7% of subjects had detectable HAI antibody
(titer 肁10), and 3%–18% had detectable MN antibody . To gain
insight into whether these low levels of antibody reflected anti-
body to H5 or were nonspecific background, we compared the
antibody response in these subjects after 1 dose of vaccine, rea-
soning that those with true H5 antibody should be primed and
have a higher response to vaccine. The day 28 GMT after dose 1
in subjects shown to have preexisting antibody to the H5N1 vi-
rus by HAI assay was significantly higher than that for subjects
without detectable antibody (56.3 vs. 19.3; P ⬍ .001, Wilcoxon
rank test) (table 5), but the percentage with a 4-fold increase in
titer did not differ significantly (18% vs. 11%; P ⫽ .34, Fisher
exact test) (table 6). When the day 28 MN titers were compared,
the subjects with preexisting neutralizing antibody had a higher
GMT than subjects without preexisting antibody (95.0 vs. 16.8;
P ⬍ .001, Wilcoxon rank test), and they were more likely to
develop a 4-fold increase in antibody level after 1 dose (30% vs.
8%; P ⫽ .001, Fisher exact test) (table 6).
The possibility that these differences were the result of an un-
equal distribution within vaccine groups was investigated. There
were no significant differences in the distribution of subjects
with preexisting antibody across the 9 groups for either HAI
(P ⫽ .37) or MN (P ⫽ .37). There were also no significant dif-
ferences in the baseline GMTs across groups for HAI (P ⫽ .28)
and MN (P ⫽ .71).
674 ● JID 2008:197 (1 March) ● Bernstein et al.
DISCUSSION
All the H5N1 vaccine and adjuvant combinations were generally
well tolerated, although local symptoms, especially pain and ten-
derness, were common. The percentage of subjects who devel-
oped pain and/or tenderness increased as the antigen content of
the vaccine increased, and the percentage was increased at equiv-
alent doses when combined with an adjuvant. The addition of
Pa aluminum hydroxide increased the frequency of pain, compared
with vaccine alone, and MF59-adjuvanted vaccines further in-
.339 creased the percentage of subjects who experienced pain. It is,
.001 however, important to note that the pain was most often as-
sessed as mild to moderate and did not lead to the withdrawal of
any subject before the second dose was administered. An in-
crease in local reactions to vaccine with MF59 has been noted
previously when this adjuvant was used with other avian influ-
enza vaccines [7, 8], as well as with seasonal influenza vaccines
[13, 14].
The antibody response to the vaccines appeared to be related
to both the quantity of vaccine and the inclusion of MF59, but
not to the inclusion of aluminum hydroxide. The antibody re-
sponse to vaccine tended to increase as quantities of vaccine in-
creased. Without an adjuvant, the percentage of subjects who
developed neutralizing antibody at a titer of 肁40 after 2 doses of
vaccine increased from 28% in the group that received 15 ␮g of
vaccine to 48% in the group that received 45 ␮g. The addition of
aluminum hydroxide did not increase the antibody response in
any of the 3 groups in which it was evaluated (i.e., the groups that
received 7.5, 15, or 30 ␮g of vaccine). In a previous report of an
H5N1 vaccine, aluminum hydroxide improved the antibody re-
sponse to a 30-␮g dose of vaccine but did not increase the anti-
body response to lower doses of vaccine (7.5 or 15 ␮g) [4]. In
fact, although the HAI GMT to the 30-␮g dose was increased by
adjuvant, the effect was not described as statistically significant,
and in the group that received a 7.5-␮g dose, the HAI GMT was
similarly decreased by the addition of aluminum hydroxide.
Thus, it does not appear that the addition of aluminum hydrox-
ide will improve the immunogenicity of subunit avian influenza
vaccines.
The addition of MF59, however, increased both the GMT
and the percentage of subjects who developed an antibody
titer 肁40. The group receiving the highest dose of vaccine (15
␮g) tested with MF59 was the only group in which ⬎50% of
the subjects were shown to have developed an antibody titer
of 肁40, by either HAI or neutralization assay. The neutraliz-
ing antibody GMT in this group was 63.0, compared with a
GMT of 15.8 in the group that received 15 ␮g of vaccine
alone. The antibody response in the group receiving 15 ␮g of
vaccine with MF59 was higher than that in the group receiv-
ing 7.5 ␮g of vaccine with MF59, but it is important to note
that the latter group received only half the dose of MF59 that
the former group received. The increased antibody response
observed here is consistent with previous reports that dem-
onstrate increased antibody titers following vaccination with
other A/H5N3 and A/H9N2 avian influenza vaccines [7, 8] as
well as seasonal influenza vaccines [13, 14]. Further evalua-
tions that use lower doses of vaccine and consistent doses of
MF59 are needed to define the optimum combination for
vaccination. Extended studies involving the elderly individu-
als and young people are also needed to determine whether
the addition of MF59 would improve the immunogencity of
A/H5 vaccines in these groups.
HAI, and more often MN, antibodies to H5N1 virus were
occasionally detected prior to vaccination. Previous studies have
also reported a low level of preexisting antibody [5, 4], without
apparent exposure to H5N1 viruses. It is unclear whether this
represents a true response or was a nonspecific reaction. Neu-
tralizing antibody may be present due to neuraminidase specific
antibody, as N1 viruses (H1N1) have circulated and have been
part of routine influenza vaccines for some time. It is also possi-
ble that this antibody represents heterosubtypic immunity from
cross-reacting epitopes from H1, H2, or H3 viruses. It is, there-
fore, interesting to note that there was some evidence that sub-
jects with preexisting antibody responded better to the first dose
of vaccine than subjects without preexisting antibody, which im-
plies that the initial antibody measured may have been H5 spe-
cific and that the improved responses were due to a booster re-
sponse, as opposed to a primary response.
In summary, 2 doses of A/H5N1 vaccine at antigen concen-
trations ranging from 7.5 to 45 ␮g per dose were well tolerated.
The addition of aluminum hydroxide did not improve antibody
response for any antigen dose tested. However, when vaccine
antigen was combined with MF59, both HAI and MN antibody
responses were substantially increased, so that titers were higher
for the group that received 15 ␮g of vaccine with MF59, com-
pared with titers for the group that received the highest dose
tested, 45 ␮g of vaccine alone. The target HAI antibody titer of
肁40 was achieved in 63% of the subjects, and the target MN
response was detected in 81% of subjects who received 15 ␮g of
vaccine with MF59, the highest dose tested with MF59. Antibody
responses were thus as high or higher than those reported pre-
viously in studies that used 45–90 ␮g of A/H5 antigen [5].
References
1. World Health and Organization. Epidemic and pandemic alert and re-
sponse (EPR), avian influenza. Available at: http://www.who.int/csr/
disease/avian_influenza/en/.
2. Tiensin T, Chaitaweesub P, Songserm T, et al. Highly pathogenic avian
influenza H5N1, Thailand, 2004. Emerg Infect Dis 2005; 11:1664 –72.
3. Tran TH, Nguyen TL, Nguyen TD, et al. Avian influenza A (H5N1) in 10
patients in Vietnam. N Engl J Med 2004; 350:1179 – 88.
4. Bresson JL, Perronne C, Launay O, et al. Safety and immunogenicity of
an inactivated split-virion influenza A/Vietnam/1194/2004 (H5N1) vac-
cine: phase I randomised trial. Lancet 2006; 367:1657– 64.
5. Treanor JJ, Campbell JD, Zangwill KM, Rowe T, Wolff M. Safety and
immunogenicity of an inactivated subvirion influenza A (H5N1) vac-
cine. N Engl J Med 2006; 354:1343–51.
6. Treanor JJ, Wilkinson BE, Masseoud F, et al. Safety and immunogenicity
of a recombinant hemagglutinin vaccine for H5 influenza in humans.
Vaccine 2001; 19:1732–7.
7. Atmar RL, Keitel WA, Patel SM, et al. Safety and immunogenicity of
nonadjuvanted and MF59-adjuvanted influenza A/H9N2 vaccine prep-
arations. Clin Infect Dis 2006; 43:1135– 42.
8. Nicholson KG, Colegate AE, Podda A, et al. Safety and antigenicity of
non-adjuvanted and MF59-adjuvanted influenza A/Duck/Singapore/97
(H5N3) vaccine: a randomised trial of two potential vaccines against
H5N1 influenza. Lancet 2001; 357:1937– 43.
9. Podda A, Del Giudice G. MF59-adjuvanted vaccines: increased immu-
nogenicity with an optimal safety profile. Expert Rev Vaccines 2003; 2:
197–203.
10. Wood JM, Robertson JS. From lethal virus to life-saving vaccine: devel-
oping inactivated vaccines for pandemic influenza. Nat Rev Microbiol
2004; 2:842–7.
11. Rowe T, Abernathy RA, Hu-Primmer J, et al. Detection of antibody to
avian influenza A (H5N1) virus in human serum by using a combination
of serologic assays. J Clin Microbiol 1999; 37:937– 43.
12. Stephenson I, Wood JM, Nicholson KG, Charlett A and Zambon MC.
Detection of anti-H5 responses in human sera by HI using horse eryth-
rocytes following MF59-adjuvanted influenza A/Duck/Singapore/97
vaccine. Virus Res 2004; 103:91–5.
13. Frey S, Poland G, Percell S and Podda A. Comparison of the safety,
tolerability, and immunogenicity of a MF59-adjuvanted influenza vac-
cine and a non-adjuvanted influenza vaccine in non-elderly adults. Vac-
cine 2003; 21:4234 –7.
14. Minutello M, Senatore F, Cecchinelli G, et al. Safety and immunogenic-
ity of an inactivated subunit influenza virus vaccine combined with
MF59 adjuvant emulsion in elderly subjects, immunized for three con-
secutive influenza seasons. Vaccine 1999; 17:99 –104.
Evaluation of Influenza H5N1 Vaccine ● JID 2008:197 (1 March) ● 675