Safety and Immunogenicity of Trivalent Inactivated Influenza Vaccine in Infants

Abstract

Background

Trivalent inactivated influenza vaccine (TIV) is not licensed for use in infants <6 months old, the group with the highest influenza hospitalization rates among children.

Methods

In this prospective, open-label study, 2 doses of TIV were administered to healthy infants aged 10–22 weeks. Adverse reactions were assessed, and hemagglutination inhibition (HAI) antibody titers were determined. Weekly telephone surveillance for influenza-like illness was conducted during the influenza season.

Results

A total of 42 infants were enrolled and completed the study. Mild local and systemic reactions were noted. In the first season (2004–2005), postvaccination HAI titers >1:32 were noted for 31.6%, 47.4%, and 21.1% of 19 subjects for H1N1, H3N2, and B strains included in the vaccine, respectively. In the second season (2005–2006), postvaccination HAI titers >1:32 were seen in 45.5%, 59.1%, and 0% of 23 subjects for H1N1, H3N2, and B strains included in the vaccine, respectively. Infants who were seronegative before vaccination (titers <1:8) were significantly more likely to have a 4-fold rise in antibody titer after vaccination, compared with infants who had prevaccination titers >1:8 (P < .001).

Conclusion

Two doses of TIV were found to be safe and moderately immunogenic against some influenza strains. The presence of preexisting maternally derived antibody was associated with significantly lower seroresponse rates to vaccination. Whether vaccination with TIV will prevent influenza in these young children remains to be determined.

Influenza is an important cause of morbidity and mortality among both children and adults. Influenza A and/or B viruses cause yearly epidemics in the United States, with an average of 36,000 deaths and 114,000 hospitalizations each year [1]. Children have the highest rates of infection, and elderly adults have the highest mortality rates [2]. Influenza is also associated with a substantial number of hospitalizations among young infants [3]. Researchers at Vanderbilt University, using a survey of the Tennessee Medicaid database, reported that hospitalization rates for infants <6 months of age were much higher than those for older children (104 hospitalizations for influenza per 10,000 infants <6 months of age versus 4 hospitalizations per 10,000 children 5–15 years of age), approaching the hospitalization rates for adults >65 years of age [4]. In addition, a recent prospective surveillance study confirmed the earlier findings, with the average annual rates of hospitalization attributable to influenza reported as 45 hospitalizations per 10,000 infants 0–5 months of age, 9 hospitalizations per 10,000 children 6–23 months of age, and 3 hospitalizations per 10,000 children 24–59 months of age [5]. In addition to the hospital burden, the same study found that, for infants 0–5 months of age, the annual rates of outpatient visits attributable to influenza were ~10-fold higher than hospitalization rates [5].

For many years the trivalent inactivated influenza vaccine (TIV) was recommended only for children with high-risk medical conditions. However, in 2004–2005, this recommendation was extended to include all children 6–23 months of age, and in 2006–2007, it was further extended to include all children 6–59 months of age [6, 7]. Because of the burden of influenza in infants <6 months of age [4, 5, 8, 9], we sought to evaluate the safety and immunogenicity of TIV when administered to infants 10–22 weeks of age.

METHODS

Study design

This was a phase 1, prospective, open-label safety and immunogenicity study of 2 doses of TIV administered to infants 10–22 weeks of age. No control group was included. The infants were also followed up with weekly telephone calls to parents or guardians during the influenza season to evaluate for influenza-like disease. The study was conducted at Vanderbilt University Medical Center (October 2004 through April 2005 and September 2005 through April 2006) and at the Cincinnati Children’s Hospital Medical Center (CCHMC) (November 2005 through April 2006). Approval to conduct the study was obtained from institutional review boards at both institutions.

Subjects

Healthy infants 10–22 weeks of age who were available for the entire study period and whose parents or guardians provided consent were eligible to participate. Patients were recruited at Vanderbilt from the Vanderbilt Primary Care Clinic and 2 additional private pediatric practices (Rivergate Pediatrics and Franklin Pediatrics Associates). At CCHMC all subjects were enrolled from a large private pediatric practice (Pediatric Associates). Exclusion criteria excluded children who (1) were born at <37 weeks of gestation; (2) had a history of hypersensitivity to eggs or egg protein; (3) had a history of wheezing or use of a bronchodilator; (4) had an underlying chronic illness (e.g., a congenital heart defect or bronchopulmonary dysplasia [BPD]); (5) had an underlying immunodeficiency or were receiving immunosuppressive therapy; (6) had participated in a clinical trial for an investigational drug or vaccine since birth; (7) had received blood products in the last 3 months; (8) had had a rectal temperature >38.0°C or an acute illness within 48 h of vaccination; (9) had been hospitalized previously, other than at birth; or (10) had received routine infant vaccinations within 14 days before influenza vaccination.

Vaccine

The standard trivalent inactivated influenza split-virus vaccine (Sanofi Pasteur) contained 15 μg of hemagglutinin for each influenza strain included in the vaccine, for a total of 45 μg per 0.5-mL dose. The vaccine formulation for the 2004–2005 influenza season was A/New Caledonia/20/99 (H1N1), A/Wyoming/3/2003 (an A/Fujian/411/2002 [H3N2]–like virus), and B/Jiangsu/10/2003 (a B/Shanghai/361/2002-like virus), and for the 2005–2006 season, A/New Caledonia/20/99/IVR-116 (H1N1), A/New York/55/2004/X-157 (H3N2, an A/California/7/2004-like strain), and B/Jiangsu/10/2003. Each subject received 0.25 mL of vaccine, injected intramuscularly in the anterolateral part of the right thigh with a 25-gauge, 58-in needle. The subjects were observed closely for ≥30 min after vaccination. The 2 doses of TIV were separated from other routine pediatric vaccines by a minimum of 2 weeks and from each other by 28 (+7) days.

Safety evaluation

Parents or guardians were asked to record local reactions (pain, swelling, or redness at the injection site), systemic adverse events (elevated temperature, irritability, drowsiness, or loss of appetite), and concomitant medications on worksheets from day 0 to day 7 after vaccination. Follow-up telephone calls were made by the study staff on days 1–3 and on days 8–10 after vaccination and again at 6 months after vaccination. Unsolicited adverse events were collected for 28 days after vaccination and serious events for 6 months after vaccination. Elevated temperature and redness and swelling at the injection site were graded on a scale of 0–4 (table 1).

Table 1.

Grading scale for elevated temperature and redness and/or limb swelling at the injection site.

Grade	Elevated temperature, °C	Injection site redness and limb swelling, mm
0	<37.5	Absent
1	37.5 to <38	<5
2	38 to <38.5	5 to 20
3	38.5 to 39.5	21 to 35
4	≥39.5	≥35

Immunogenicity evaluation

Serum samples were obtained before administration of the first dose of TIV and 28 days after administration of the second dose of TIV; they were stored at −20°C until analyzed at Vanderbilt University. Hemagglutination inhibition (HAI) antibody titers for both influenza A and B vaccine antigens were determined in duplicate, with paired specimens tested simultaneously. The antigens for each year were supplied by the Centers for Disease Control (CDC). Serum samples were treated with receptor-destroying enzyme and then heated to 56°C for 30 min. HAI assays were performed at a starting dilution of 1:8, with subsequent serial 2-fold dilutions. If HAI titers were <1:8, they were assigned a value of 1:4 dilution. A seroprotective titer for HAI was defined as ≥1:32 [10]. Seroconversion was defined as a ≥4-fold rise in HAI antibody titers 1 month after the second TIV dose, in comparison with prevaccination titers.

Microneutralization (MN) assays [11] were performed at Vanderbilt only for influenza B, according to the CDC protocol (provided by Jacqueline Katz). Assays began at a starting dilution of 1:10, with subsequent serial 2-fold dilutions. If MN titers were <1:10, they were assigned a value of 1:5. A seroprotective titer for MN was defined as ≥1:40 [10]. Seroconversion by MN was defined as a ≥4-fold rise in antibody titer 1 month after the second TIV dose, in comparison with prevaccination titers.

The HAI antigens were those received from the CDC identification kit for that year, and the influenza B virus for the MN assays was received and passaged in eggs to a titer of 10⁷–10⁸. Specifically, for 2004–2005, HAI assays included A/New Caledonia/20/99, A/Wyoming/3/2003, B/Shanghai/361/2002 (ether extracted), and B/Hong Kong/330/2001 (ether extracted); for 2005–2006, they included A/New Caledonia/20/99, A/California/7/2004, and B/Shanghai/361/2002 (ether extracted). For MN, B/Hong Kong/1434/2002 was used in 2004–2005, and B/Hong Kong/1434/2002 was used in 2005–2006. The HAI assays for influenza B virus were done with ether-extracted split virus. The negative controls for both assays came from a pool of serum samples obtained from 15 infants 8–12 months of age who were predicted to be seronegative by virtue of age and lack of exposure to influenza, and all samples were confirmed to be seronegative by HAI and MN assay.

Monitoring for influenza disease

The beginning of influenza season was defined as the date when ≥2 influenza-positive respiratory cultures or polymerase chain reaction (PCR) test results were obtained for 2 consecutive weeks in the clinical or research laboratories at Vanderbilt or Cincinnati. The end of the influenza season was defined as the date when <2 influenza positive respiratory cultures or PCR tests were obtained for 2 consecutive weeks. Infants were closely monitored during the influenza season [5]. During the 2004–2005 influenza season, all children with any respiratory symptoms were seen, and nasal wash samples were obtained for rapid influenza testing and culture, confirmed by PCR when there was a discrepancy. During the 2005–2006 season, infants were seen only for specific lower respiratory findings, such as wheezing, shortness of breath, or chest congestion, and for rectal temperatures ≥38.3°C.

Statistical analysis

The frequencies of individual local or systemic adverse events were summarized and classified for both doses 1 and 2 in predefined grades for 0–7 days after vaccination. The percentages of individuals who achieved seroprotection (>1:32 for HAI or >1:40 for MN) and seroconversion (≥4-fold rise in antibody titer after vaccination) were reported for years 1 and 2 and compared using the χ² or Fisher’s exact test for each strain. The geometric mean titer (GMT) ratios were calculated as the ratios of postvaccination titers to prevaccination titers at geometric mean scale. The confidence intervals (CIs) of GMT ratios were calculated on the basis of Gaussian approximation of the natural logarithmic transformation. Seroconversion rates were compared between infants with no preexisting titers and those with maternally derived preexisting antibody titers for each strain, by use of Fisher’s exact test. The decay of maternal antibody over time was taken into account. All analyses used a 2-sided significance level of .05 and were performed with SAS software (version 9.1; SAS Institute) and R software (version 2.4.1.; http://www.r-project.org).

RESULTS

Subjects

During the 2 influenza study seasons, 457 parents or guardians of eligible infants were approached about the study, and 42 agreed to enroll. The main reason for parental refusal was the frequency of the blood sampling required by the study. During the first influenza season (2004–2005), 19 infants were enrolled at the Vanderbilt University Medical Center, and during the second season (2005–2006), 13 infants were enrolled at Vanderbilt and 10 infants were enrolled at CCHMC. The mean, median, and range for the ages of the infants were 14.4 weeks, 13.9 weeks, and 10.1–21.9 weeks, respectively, in 2004–2005; in 2005–2006, the mean, median, and range were 17.2 weeks, 18.3 weeks, and 10.0–22.6 weeks. Of the 42 children enrolled, 36 were white non-Hispanic, 1 was white Hispanic, 2 were black, and 3 were multiracial; 20 (48%) were male.

Safety data

The majority of systemic reactions in the 7 days after vaccination were grades 1 or 2 (figure 1), for both vaccine doses. After doses 1 and 2, a total of 0% and 7% of subjects, respectively, had a rectal temperature of ≥38°C. One subject who had an elevated rectal temperature of 39.1°C on days 4–7 after vaccination was diagnosed with otitis media. No severe adverse events related to vaccination were reported. Most of the local reactions were grade 1 or 2 and occurred on day 0 or on day 1 after vaccination (figure 2). In addition, most local and systemic symptoms, except for elevated temperature, were observed during days 0–3, rather than during days 4–7 (P =.002) (table 2).

Figure 1.

Percentages of subjects with systemic reactions after dose 1 and dose 2 of trivalent inactivated influenza vaccine, by grade of reaction. Data were collected for 7 days after vaccination; day 0 is the day of vaccination.

Figure 2.

Percentages of subjects with local reactions after dose 1 and dose 2 of trivalent inactivated influenza vaccine, by reaction grade. Data were collected for 7 days after vaccination; day 0 is the day of vaccination.

Table 2.

Summary of reactogenicity, by days after vaccination.

Reaction	Days 0–3	Days 4–7	Pa
Elevated rectal temperature	17 (40.5)	13 (31.0)	.495
Irritability and/or fussiness	27 (64.3)	12 (28.6)	.002
Drowsiness	20 (47.6)	3 (7.1)	<.001
Loss of appetite	13 (31.0)	4 (9.5)	.028
Any systemic symptoms	35 (83.3)	19 (45.2)	.001
Pain	9 (21.4)	0 (0.0)	.002
Redness	8 (19.0)	0 (0.0)	.005
Swelling	3 (7.1)	0 (0.0)	.241
Any local symptoms	15 (35.7)	0 (0.0)	<.001
Any symptoms	35 (83.3)	19 (45.2)	.001

NOTE. Data are no. (%) of patients, unless otherwise indicated (n =42).

^aFisher’s exact test.

Immunogenicity data

Complete immunogenicity data were available for H1N1 and B strains for 40 of 42 subjects; data for H3N2 was available for 39 of 42 subjects. For 2 subjects insufficient serum was obtained for any assays, and 1 subject missed a visit at which a serum sample was to be obtained. HAI titers for the H1N1, H3N2, and B strains included in the vaccine were ≥1:32 in 31.6%, 47.4%, and 21.1% of subjects, respectively, during season 1; during season 2, HAI titers were ≥1:32 in 45.5%, 59.1%, and 0% of subjects, respectively (table 3). Titers for the H1N1, H3N2, and B strains included in the vaccine increased 4-fold in 42.1%, 38.9%, and 5.3% of subjects, respectively, in season 1 and in 42.9%, 66.7%, and 0% of subjects in season 2 (table 3). Finally, GMT ratios were similar for H1N1 and B during both years but higher for H3N2 during year 2 (table 3). GMTs after vaccination were similar for H1N1 during both years but were higher for H3N2 and lower for B during year 2 (table 3). The major circulating virus for 2004–2005 was A/California/7/2004 (H3N2), a drift strain from the vaccine strain. HAI antibody titers ≥1:32 for A/California were seen in 5.3% of vaccinees (1 of 19), and 5.6% (1 of 18) had a 4-fold rise from baseline titers. The HAI GMT was 5.77 (95% CI, 4.35–9.63), and the GMT ratio was 1.17 (95% CI, 0.86–1.58).

Table 3.

Immunogenicity data for 2004–2005 and 2005–2006.

Assessment	Influenza strain
	H1N1	H3N2	B
	Subjects with HAI titers ≥1:32, % (proportion) of subjects
2004–2005	31.6 (6/19)	47.4 (9/19)	21.1 (4/19)
2005–2006	45.5 (10/22)	59.1 (13/22)	0 (0/22)
Pa	.36	.45	.04
Subjects with a 4-fold increase from baseline titers, % (proportion) of subjects
2004–2005	42.1 (8/19)	38.9 (7/18)	5.3 (1/19)
2005–2006	42.9 (9/21)	66.7 (14/21)	0 (0/21)
Pa	.96	.08	.48
GMT ratio (95% CI)
2004–2005	2.23 (1.17–4.24)	1.71 (0.67–4.40)	1.0 (0.66–1.52)
2005–2006	2.07 (1.02–4.20)	8.83 (4.15–18.80)	0.97 (0.86–1.09)
GMT after vaccination (95% CI)
2004–2005	14.87 (9.41–23.51)	21.42 (12.23–37.52)	8.30 (5.35–12.86)
2005–2006	16.00 (9.28–27.59)	39.90 (20.77–76.64)	4.13 (3.87–4.41)

NOTE. CI, confidence interval; GMT, geometric mean titer; GMT ratio, ratio of post- to prevaccination GMTs; HAI, hemagglutination inhibition.

^aP values for comparison between seasons.

Seronegative infants (those with prevaccination titers <1:8) were significantly more likely to have a seroresponse after the second dose of TIV than seropositive infants (those with prevaccination titers ≥1:8), for both H1N1 (P <.001) and H3N2 (P <.001) (table 4). Because HAI titers have typically been low for B antigens, MN assays were done to confirm the response to B antigens [12, 13]. In our study, immune responses to influenza B were uniformly poor when measured by both HAI and MN assays. As measured by MN assays only, 1 infant had seroprotective titers for influenza B after the second dose of TIV.

Table 4.

Subjects with a 4-fold increase in hemagglutination inhibition titers, in comparison with prevaccination titers.

Prevaccination titer	Subjects with >4-fold response, %	GMT (95% CI)
H3N2 strain
<1:8 (n =23)	82.6a	53.41 (28.89–98.74)c
≥1:8 (n =16)	12.5a	12.88 (8.46–19.62)c
P	<.001	<.001
H1N1 strain
<1:8 (n =23)	65.2a	20.99 (12.93–34.07)c
≥1:8 (n =17)	11.8a	11.09 (6.79–18.09)c
P	<.001	.06
B strain
<1:8 (n =30)	3.3b	4.81 (3.93–5.90)c
≥1:8 (n =10)	0b	9.85 (5.07–19.12)c
P	.99	.04

NOTE. CI, confidence interval; GMT, geometric mean titer.

^aχ² test.

^bFisher’s exact test.

^cTwo-sample t test, assuming unequal variance.

Influenza surveillance

During the first influenza season, the 19 children enrolled in the study had 27 ill visits. Two children had laboratory-confirmed influenza disease. One child had culture-confirmed influenza B with a mild clinical course characterized by low-grade fever for 1 day and diagnosis of acute otitis media. This subject had both prevaccination and postvaccination titers of 1:16 for influenza B. One child had a positive rapid diagnostic test result for influenza A and a negative culture but a positive PCR result for influenza A. This child had a runny nose and fever for 2 days and was diagnosed with a URI. This subject had both prevaccination and postvaccination titers of <1:8 for both H1N1 and H3N2. During the second influenza season, the 23 children enrolled in the study had 12 ill visits, and none had culture-confirmed or PCR-confirmed influenza. For 22 of 42 subjects (52.4%), ≥1 parent or guardian had received the influenza vaccine. For 5 of the 20 subjects whose parent(s) or guardian(s) did not receive the influenza vaccine, ≥1 sibling had received the vaccine.

DISCUSSION

Although infants <6 months of age have the highest hospitalization rates for influenza in the pediatric population, neither influenza vaccines nor antiviral medications are approved for this age group [4]. Therefore, these infants are susceptible to influenza and have limited prevention and treatment options. This pilot study demonstrates that TIV was well tolerated in this vulnerable population.

To our knowledge, there has been only 1 study of TIV use in infants <6 months of age published previously [14]. In that study, high-risk children with BPD or congenital heart disease who were aged 3–5 months or 6–18 months received 2 doses of TIV administered 4–6 weeks apart without other concomitant vaccines. No placebo group was included. Seven of the 62 infants aged 3–5 months had mild local reactions during the first vaccination, but none had reactions after the second vaccination. Similar to our findings, no children aged 3–5 months had fever after either dose of TIV. Of the 51 children aged 6–18 months, only 2 experienced adverse events. One child had mild redness and tenderness at the injection site accompanied by vomiting and irritability after the first injection, but tolerated the second injection without incident. The second child was a 13-month-old with BPD and no history of seizures; this child had a grand mal seizure 24 h after the first injection, associated with a fever of 39.5°C. No source was discovered for the fever, and the second influenza vaccination was not given. The safety and reactogenicity profiles of TIV in both studies are consistent with other reports for older infants and children [15, 16].

In our study, the serological responses to H1N1 and H3N2 during the 2005–2006 season were numerically higher than those during the 2004–2005 season, but not significantly so. The responses to B were uniformly low; only one-fifth of the infants in 2004–2005 and none in 2005–2006 had an HAI titer ≥1:32. The lower responses to B antigens are consistent with previous reports [12, 13]. A higher percentage of infants had titers ≥1:32 for H3N2, compared with titers for H1N1 or B, and this variability in response to the different influenza antigens has been reported elsewhere [14, 17]. For example, in the study by Groothuis et al., which spanned several influenza seasons, 14 (93%) of 15 infants aged 3–5 months developed protective titers (defined as >1:32) for A/Mississippi/11/85 (H3N2) after the second vaccine dose [14]. However, during other influenza seasons, the response rates to H3N2 vaccines varied. With respect to other H3N2 antigens, only 11%, 28%, and 55% of the infants achieved protective titers for A/Sichuan/2/87, A/Leningrad/360/86, and A/Philippines/82, respectively (1985–1988). Similar to our findings, compared with the percentage of infants with protective titers for H3N2, the percentages of infants aged 3–5 months with protective titers were lower for H1N1 and B: 23% and 26% for A/Chile/83 (H1N1) and A/Taiwan/1/86 (H1N1), respectively; and 10%, 10%, and 27% for B/Victoria/2/87, B/Ann Arbor/1/86, and B/USSR/1/83, respectively.

The ability of preexisting maternal antibodies to suppress the serologic responses in infants who receive either live or inactivated vaccines has been demonstrated elsewhere [18–24]. In this study, we also found that infants with preexisting maternal antibodies were less likely than infants without preexisting maternal antibodies to have a 4-fold increase in antibody titers to H1N1 and H3N2 after their second dose of TIV. This suppression was not seen in the study by Groothuis et al. [14], because all infants had prevaccination titers of <1:8 to all antigens tested, which may be explained by the inclusion of many patients with BPD who were born prematurely and had little transplacental maternal antibody transfer. Because the study by Groothuis et al. [14] also demonstrated lower antibody responses to some of the influenza antigens in the absence of maternal antibody, immunologic immaturity may also have a role in the reduced responses.

Although the majority of infants who received TIV in the presence of maternal antibody failed to demonstrate significant increases in HAI antibody titers, these infants may have been primed for memory responses or generated influenza-specific T cells. These immune parameters were not assessed in this pilot study. Priming in the presence of maternal antibody was demonstrated for hepatitis A vaccines in a study by Dagan et al., in which 2 groups of infants were given hepatitis A vaccine at 2, 4, and 6 months of age [22]. Infants with preexisting maternal antibodies to hepatitis A had lower mean titers at 7 months of age than infants without preexisting maternal antibodies (508 vs. 1656 mIU/mL, respectively; P <.001), suggesting an inhibitory effect [22]. However, in a subsequent study by the same group, the infants with maternal antibodies received a fourth dose of hepatitis A vaccine at 12 months of age and were compared with infants with no preexisting maternal antibodies who received a first dose of hepatitis A vaccine at 12 months of age. The mean titers were significantly higher in those infants with preexisting maternal antibodies than in those who received their first dose of vaccine at 12 months of age (1902 vs. 120 mIU/mL, respectively; P =.001) [22]. Thus, these data suggest that maternal antibodies did not prevent T and B cell priming and subsequent memory responses, although the humoral responses obtained after the primary series were lower in infants with preexisting maternal antibodies [22]. Further studies of vaccination with TIV in young infants are needed to address definitively the question of induction of memory and cell-mediated immunity.

Our study has several limitations. It was a pilot study with a small sample size, and although we did not demonstrate any serious adverse events associated with administration of TIV, larger numbers are needed to provide added assurance of safety. It is encouraging to note that neither our study nor that of Groothuis et al. found any evidence of enhanced influenza disease after receipt of TIV, in contrast to the findings of enhanced respiratory syncytial virus (RSV) disease after administration of the formalin-inactivated RSV vaccine [14, 25]. However, because only 2 children were infected with influenza, our power to assess enhanced disease was low. In terms of immunogenicity, the only correlates of protection measured in the study were serologic, and cellular responses were not measured. These infants were followed up for only 1 influenza season, and their responses to subsequent doses of TIV are not known. Finally, the efficacy of TIV depends on variables such as the age of the individual at vaccination, the match between the vaccine strain and the circulating strains, the immunologic maturity of the individual, and the presence of preexisting antibody titers. With so few children enrolled, this study was not powered to evaluate efficacy.

The administration of TIV to infants <6 months of age appears to be safe and immunogenic, but in the presence of maternal antibodies, the percentage of infants achieving protective titers was low. Because it is not known whether these infants were primed by TIV for memory responses, additional immunologic studies are needed in subsequent studies of TIV in infants. In the meantime, it is important to encourage pregnant mothers [26, 27] and close contacts of young infants to be vaccinated, in order to protect vulnerable young children from influenza.