Half- vs Full-Dose Trivalent Inactivated Influenza Vaccine (2004-2005):  Age, Dose, and Sex Effects on Immune Responses FREE

The efficacy and cost-effectiveness of trivalent inactivated vaccine (TIV) has resulted in expanded recommendations for immunization, particularly for healthy populations involved in caregiving and first responder missions.^1- 8 Since 2002, optimum influenza vaccine delivery has been impaired as a result of supply shortages. With the abrupt loss of half the anticipated national influenza vaccine supply in October 2004, the option of using a reduced dose for immunization of healthy, high-priority groups became a critical consideration. A prior study by Treanor et al⁹ (2001-2002 season) suggested that immune response differences among subjects given a half dose of TIV were not substantially inferior in healthy subjects aged 18 to 49 years.

The need to extend the evaluation of reduced TIV doses to a different formulation resulted in the development of a rapid-response, interagency collaborative protocol implemented in November 2004 by the Walter Reed Health Care System (WRHCS) Allergy-Immunology Department and Vaccine Healthcare Center (VHC) on behalf of the Office of the Army Surgeon General. The trial was designed to replicate the noninferiority study and include healthy subjects aged 50 to 64 years. The primary outcome measure was the development of serum hemagglutination inhibition (HAI) antibody titers of 1:40 or higher to the A/Fujian/411/2002 (A/H3N2)-like, A/New Caledonia/20/99 (A/H1N1)-like, and B/Shanghai/361/2002 (B)-like antigens used in the 2004-2005 TIV. Antibody responses correlate with protection and are used as measures of immunologic comparability.¹⁰

Although not designed to be an efficacy study, the project included secondary end points of inpatient and outpatient health care encounters during the 2004-2005 influenza season using an existing medical encounter database for Military Health System (MHS) medical services both within and outside military medical facilities.

This prospective, single-blind, randomized clinical trial was approved by the Walter Reed Army Medical Center (WRAMC) Department of Clinical Investigation. Study implementation included compliance with good clinical practices as defined by published guidelines for studies involving an investigational new drug (IND #12019 held by the Office of the Army Surgeon General). The US Army Medical Materiel Development Activity served as the sponsor's representative. All subjects provided written informed consent.

The goal of the study was to provide additional information for future public health decisions concerning the best use of limited influenza vaccine supply. In that situation, moderately diminished responses shared by a large segment of the population would be preferred to maximum responses in a more restricted group, particularly since lower responses are still correlated with clinical benefit, particularly in populations receiving repeated doses.^10- 11 The study was not intended to demonstrate equivalence of antibody responses using half-dose vaccine, since diminished antibody responses have previously been described.

Healthy MHS beneficiaries in the National Capital Region of Washington, DC, were enrolled from 2 sites: Allergy-Immunology-Immunization Clinic, WRAMC, and Pentagon/DiLorenzo Health Clinic, Arlington, Virginia. There were 8 randomization strata based on sex, age group (18-49 years vs 50-64 years), and self-reported receipt of at least 1 dose of TIV within the previous 3 years. Subjects within each stratum were randomized to receive either a full dose (0.5 mL, 15-μg hemagglutinin antigen per strain) or half dose (0.25 mL, 7.5-μg hemagglutinin per strain) of licensed 2004-2005 TIV (Fluzone; Aventis Pasteur, Swiftwater, Pennsylvania). Randomization was performed through the use of envelopes sorted by age, sex, and vaccination status and assembled by McKesson Biosciences, Rockville, Maryland. Sealed envelopes were assigned as subjects were individually enrolled and were delivered to the study nurse administering the vaccine by the randomization monitor. Each envelope contained coding for either a full or half dose of TIV. The code was known to the study nurses, but subjects were blinded to vaccine dose until after completing a side effects diary and surveys related to influenza-like illness (ILI). Blinding was conducted by preventing subjects from viewing syringe volumes. Serum samples were obtained before and 21 days after (range, 19-28 days) vaccination, stored at −80°C, and batch shipped on dry ice for antibody assays. Subjects completed an online 21-day symptom diary after vaccination, which was reviewed and verified during the second visit or via telephone. Subjects also recorded oral temperatures once daily (days 0-3 after vaccine) and symptoms of fever or ILI (days 4-21). Severity of symptoms was scored using a visual analog scale (VAS) of 0 to 5, with 5 being the most severe form of pain or adverse functional impact. Solicited symptoms included injection site reactions (ie, pain, redness and/or swelling greater than 2 in (5.08 cm) in diameter, numbness and/or burning) and systemic symptoms (ie, muscle and/or joint pains, headache, other body aches and pains, swollen lymph nodes, runny nose, sore throat, cough). Immunizations and sample collection occurred between November 8 and December 23, 2004.

The Army Medical Surveillance Activity Defense Medical Surveillance System (DMSS) was used to determine if there were differences between the number of outpatient visits and inpatient days for ILI or complications among subjects receiving full- vs half-dose vaccine.¹² With the use of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes, a DMSS search strategy was implemented in January 2006 to assure that all medical encounters between enrollment and the end of the influenza season (March 31, 2005) would be captured. Diagnostic categories examined included pneumonia and ILI (ICD-9-CM codes 480-487), respiratory and circulatory events that could be complications of influenza (ICD-9-CM codes 390-519), as well as respiratory illness symptoms that might reflect less precise coding of ILI. These data were secondary outcome measures.

Subjects with a history of significant underlying illness including immunosuppressive conditions or therapies or those eligible for a full dose of influenza vaccine under the post October 4, 2004, national guidelines (at high risk, including pregnancy, or with special operational requirements such as deployment to the Middle East) were excluded from participation and were referred to the standard influenza vaccination program. Additional exclusion criteria were based on the Centers for Disease Control and Prevention (CDC) and/or Department of Defense (DoD) criteria for priority immunization that year. Inclusion criteria were based on the remaining CDC and/or DoD priority prior to the shortage announcement.¹³

According to the emergency vaccine supply shortage and public health guidelines in effect after October 5, 2004, enrolled subjects would not have been eligible for immunization; this fact was considered ethical justification for blinding subjects. Vaccination of nonpriority individuals in the study was considered ethical because the study used less than 0.01% of the DoD supply and had a potential for obtaining information that could benefit large populations in future influenza seasons.

Serum samples were coded to ensure laboratory blinding related to subjects' identities and study groups. Hemagglutination inhibition was conducted in a single laboratory using standardized methods (N.J.C., Influenza Branch, CDC).^9,14- 16 Paired serum samples from each subject were tested simultaneously.

With 640 subjects within each age stratum (18-49 or 50-64 years), the noninferiority study design provided 89% power to show that the half-dose vaccine does not cause a 20% or greater loss in the proportion of subjects with postimmunization HAI antibody titers of 1:40 or higher for each vaccine antigen, assuming the true loss is 10% (ie, proportions of 85% vs 75% for high and low dose, respectively). These margin assumptions were based on values observed by Treanor et al.⁹

Three parameters were used to assess differences in antibody responses. The primary end point was the frequency of HAI antibody titer of 1:40 or higher after immunization; secondary end points were 4-fold or greater increases in titer and geometric mean titer (GMT) of HAI antibody after immunization. Antibody titers below the detection limit of 1:10 were counted as 1:5 in calculations of GMT and fold rise. Criteria for an unacceptable “substantially diminished response” after immunization with half-dose vaccine include the following: proportion of subjects with HAI antibody titer of 1:40 or higher reduced by more than 20%; proportion of subjects with 4-fold or greater increase reduced by more than 20%; and ratio of GMTs for full vs half dose of 1:5 or higher

In each analysis, the half-dose response was concluded to be no more than moderately diminished if the 95% confidence intervals (CIs) of the dose-related differences were within the established limit of acceptability. The 95% CI for differences in proportions were calculated assuming asymptotic normality for sample proportions; CIs for ratios of GMTs were calculated by first deriving CIs for the log titers using a t distribution and then back-transforming to the original scale.

Unadjusted χ² tests were used to compare proportions; the Fisher exact test was used when expected counts were below 5. Continuous variables and titer measurements were compared using unpaired t tests adjusted for unequal variances when the F test for equality of variances had a P value lower than .05. The possibility that halving the vaccine dose affected responses differently in different groups (eg, by age or sex) was examined by assessing the interaction in analysis of variance or linear regression (for log₂ titers) and in logistic regression (for dichotomous responses). Side effects data were collapsed into 2 categories (severity on VAS ≥3 of 5 vs <3 of 5) for reporting and statistical comparisons. The nominal significance level of .05 was used throughout. No adjustments were made for multiple comparisons. Analyses were performed using SAS version 9.1 (SAS Institute, Cary, North Carolina) or Stata version 8.2 (StataCorp, College Station, Texas) statistical software.

Influenza-like illness and complications resulting in either inpatient or outpatient medical encounters were compared between dose groups (by age) by calculating the relative risk and its 95% CI. ICD-9-CM codes were grouped for upper respiratory or lower respiratory tract illness consistent with ILI or complications; analysis involved the percentage of subjects having 1 or more documented medical visit within the influenza season (through March 31, 2005 but after immunization). Separate analyses were performed for cardiovascular and other visits as well as hospitalizations.

A total of 1316 eligible subjects were enrolled in November 2004. Serologic analyses are reported for 1114 (92.6%) previously vaccinated subjects. As illustrated in Figure 1, serologic data analyses excluded subjects not previously vaccinated (n = 56 [too small for analysis]), those satisfying exclusion criteria identified after enrollment (n = 15), and/or those without evaluable paired specimens (n = 137). Of the 1316 subjects, 1259 (95.7%) completed the 3-day diary, including 1203 subjects who were previously vaccinated in the past 3 years. Table 1 summarizes the baseline characteristics of previously vaccinated subjects. Significant differences were noted between age groups in ethnicity and body mass index (P = .02 for both). The younger group had significantly higher baseline antibody levels (proportion with titers ≥1:40 GMTs) than the older group for the influenza A/H1N1 and B antigens (P < .001 for both) but not for A/H3N2. Overall, randomization yielded relative balance on baseline characteristics between the full- and half-dose groups; only GMT for A/H1N1 and ethnicity showed modest differences within the older age group (data not shown). Previously vaccinated subjects without serologic data showed no differences in baseline characteristics from the 1114 subjects with serologic data. Previously vaccinated subjects had significantly higher (P < .001 for all, data not shown) baseline HAI antibody titers for all 3 antigens (data not shown). Figure 2 illustrates the reverse cumulative distribution graphs for antibody titers within each age group, by viral strain and vaccine dose.

Table 2 summarizes the differences in antibody responses according to vaccine dose and age group. The proportions of subjects aged 18 to 49 years with titers of 1:40 or higher against influenza A/H1N1, A/H3N2, and B antigens after immunization were 11.8%, 8.3%, and 5.0% higher, respectively, among subjects given a full dose compared with those given a half dose and 15.7%, 9.5%, and 8.8% higher, respectively, among subjects aged 50- to 64 years. In both age groups, GMTs and proportion with 4-fold or higher increases in antibody titer were significantly higher (P < .05 to P < .001, see detailed P values in Table 2) in the group given a full dose, with the exception of the proportion with an HAI antibody titer of 1:40 or higher against influenza B (P = .19 in the younger age group and P = .09 for A/H1N1, a 4-fold rise in the older age group). However, when analyzing serologic end points for “substantially diminished response,” there was no evidence of substantial inferiority of half dose in subjects aged 18 to 49 years (Figure 3A), which is similar to results of Treanor et al⁹ (Figure 3B). Figure 3C illustrates the corresponding result for subjects aged 50 to 64 years. Differences due to dose appeared greater in subjects aged 50 to 64 years, but they were not significantly greater (logistic regression and analysis of variance for interaction). It is noteworthy that 90% of the 18- to 49-year age group were 35 years or older; in the 50- to 64-year age group, 81% were between the ages of 50 and 59 years, limiting the relevance of the analyses for those 60 years or older.

Table 3 provides a further breakdown of GMTs by sex. For all 3 antigens, women tended to have higher GMTs than men in the same age and dose group. The GMTs for women given a half dose were similar to the GMTs for men given a full dose. We also performed regression analyses of log postimmunization titers that adjusted for age group, dose level, sex, and preimmunization antibody titers. These analyses showed that the difference attributable to female sex equaled or exceeded that between the full- and half-dose vaccine. Similar patterns were observed when analyzing the effects of the same factors on postimmunization titer of 1:40 or higher and 4-fold or greater rise in titer (data not shown). When not adjusting for preexisting antibody, the sex effect was again at least similar to the dosage effect (data not shown).

A total of 1259 subjects completed the day 0 to day 3 postimmunization side effects diary. Injection site reactions were more frequent in the full-dose group for redness and/or swelling greater than 2 in (5.08 cm) in diameter (13.4% vs 8.6%; P = .006). Although injection site pain was greater for full vs half dose (19.9% vs 14.4%; P = .01), when analyzed for clinically significant pain levels (VAS, ≥3 of 5), significant sex- but not dose-dependent pain differences were identified. Table 4 summarizes dose and sex comparisons of clinically significant side effects, symptoms of ILI, and medical events (eg, hospitalizations and unscheduled medical visits) during the serum collection period. Joint and/or muscle pain (VAS, ≥3 of 5) were significantly different (P = .02 and P = .03, respectively) for dose and sex; headache, other pains, impact of fatigue, and overall side effects were significantly greater in women (P < .001, P = .002, P = .02, and P = .05, respectively). No other adverse event differed significantly by dose, and there was no evidence of ILI symptom differences by sex or dose during the 21 days after immunizations. There were 3 hospitalizations for medically unrelated events, with 1 prior to the postvaccination serum collection resulting in exclusion from serologic analyses.

For the period of November 1, 2004, through March 31, 2005, analysis of the DMSS search for ILI and complications identified no significant differences in the frequency of respiratory illness–associated, cardiovascular, or unrelated medical visits. Visits for mental health and genitourinary problems were not reviewed. There were no deaths in the study population during the study. Emergency department visits cannot be distinguished from outpatient clinic visits in the DMSS database. No other significant adverse events related to immunization were identified during the follow-up period.

Table 5 summarizes analyses of outcomes based on the number of medical visits documented within each age group by vaccine dose, for previously vaccinated participants. There were no significant differences between vaccine doses for either age group or between sex subsets. The 95% CIs show that for half vs full dose, true ILI rates are unlikely to be more than 1.46-fold higher among the younger group and 2.2-fold higher among the older group. The substantially lower total number of visits documented for the subjects aged 50 to 64 years may reflect that the group includes predominantly nonactive duty beneficiaries who have a lower priority for primary care appointments within the MHS and who often have alternative health insurance for which medical visits are not recorded within the DMSS.

The results of this study are consistent with a similar report for the 2001-2002 influenza vaccine.⁹ Results from the 2 studies using different TIV formulations confirm that substantially inferior immune responses following 50% vaccine dose reduction in healthy adults younger than 50 years would be unlikely. The study by Kramer et al¹¹ with a half-dose influenza vaccine, although with a smaller population, is consistent with the conclusions of this study. Given the benefits of immunizing healthy working adults⁴ and caregivers,^14- 15 these data support the validity of a dose reduction strategy in the setting of vaccine shortages.¹⁶ Among subjects aged 50 to 64 years (reflecting predominantly ages 50-59 years), the primary immunologic outcome following half-dose vaccination was nearly as good as for the younger group; however, observed GMTs were at the border of the acceptable range, with some 95% CIs including unacceptable values.

Antibody responses among women given a half dose of TIV were similar to or greater in magnitude when compared with responses among men given a full dose. These findings suggest that guidelines for vaccine use during shortages should take sex as well as age into consideration. Although data on sex differences in immune responses to vaccines are limited,¹⁶ women have higher absolute CD4⁺ lymphocyte counts¹⁸ and production of T_H1 cytokines after immunization¹⁹ as well as more sustained responses to antigenic challenge.²⁰

Our data support the use of a half dose of TIV in healthy persons up to age 50 years based on serum antibody responses; data for the older age group suggest similar findings and merit further study. Reduced dosing could have a significant impact on the response to vaccine shortages, particularly at a local level when faced with considerable delays in vaccine supply delivery. In view of the trend toward reduced side effects, reducing the dose may improve vaccine acceptability in some populations. Studies related to reduced vaccine dosing and optimal criteria (eg, sex, age ranges) are limited; further investigation is merited, particularly in those up to age 60 years and with increased focus on relevant differences between previously vaccinated and unvaccinated subjects.¹⁶

In the present study, the overall response to the A/H1N1 component of the vaccine was rather weak and lower than that reported by Treanor et al.⁹ However, antibody avidity and potential efficacy may increase after immunization, even in the absence of increases in antibody levels.¹⁰ Some recent publications suggest that A/H1N1 strains may be less well covered by the TIV in some seasons.^21- 24 However, we observed no evidence of increased ILI outcomes in the half- vs full-dose groups, regardless of age group. Public health surveillance did not detect evidence of vaccine failure, as described for the 2003-2004 influenza vaccine season.²⁴ These differences may also be explained by the fact that most of our subjects had received 2 to 3 doses of TIV in the 3-year period before the study, whereas a high proportion of subjects in the study by Treanor et al⁹ had been immunized at most once in recent years (oral communication, J.J.T., 2007). In addition, the subjects in the study by Treanor et al⁹ were younger, with a mean age of 33.5 years vs 42.3 years for our 18- to 49-year age group; 90.0% of our subjects were 35 years or older. Based on influenza season activity reports, there was no locally increased influenza activity until January 2005, well past the end of serum sample collections on December 23, 2004, and therefore unlikely to have affected study results (http://www.cdc.gov/flu/weekly/fluactivity.htm).²⁵ Another factor that might affect immune responses is body mass index. There was a mild association of higher antibody responses with higher body mass index, but multivariate analyses showed that this effect does not explain or reduce the importance of the sex, dose, or age effects. The sex factor in relation to more local reactions and higher antibody responses to influenza vaccine has been previously described but has not been sufficiently considered in the context of vaccine supply shortages.^26- 28

Our observed relative risks of ILI for the half- vs full-dose vaccine are very close to 1.00 (no difference), but the 95% CIs include a fairly wide range of possible increased risk. Directly assessing modest differences in risk by dose would clearly require very large and costly studies. A recent review of influenza vaccine trials in healthy adults by the Cochrane Collaboration²⁹ concluded that only 30% of ILI is prevented by immunization in adults. Thus, particularly when faced with a vaccine supply crisis, it is sensible to rely on serological results to guide public policy regarding dosing. This was the origin of our primary objective for assessing immunologic noninferiority with a different vaccine in a different season. Our results substantiate that the observations by Treanor et al⁹ of the similarity between the half- and full-dose parenteral vaccine are potentially applicable to future vaccines. On an immunologic basis, the principle of a wide dose response in healthy individuals should be applicable to any protein vaccine construct, particularly the recombinant hemagglutinin influenza vaccine.³⁰

Although previous reports describe the use of reduced TIV doses, most involve alternative means of administration (intradermal, transdermal, or by needle-free jet injectors) not in common use and potentially requiring new training and skill sets.^16,31- 34 The present study represents the second largest study, to our knowledge, on half- vs full-dose TIV in a different season with a different vaccine formulation, supporting a reduced vaccine dosing strategy in healthy persons younger than 50 years.^9,16 Our data suggest that this approach could be extended beyond age 49 years, particularly for healthy women, but further studies are required. Half-dose TIV represents a viable alternative strategy for managing supply shortages to optimize timely delivery, particularly to critical personnel such as first responders and service members. The fact that such a strategy also reduces side effects may have the added benefit of improved overall vaccine acceptability, particularly in subpopulations experiencing more severe adverse effects. As recommendations for influenza immunization expand and evidence that elderly persons (men older than 60 years) may require higher doses of vaccine for optimal responses,^28,35 reduced doses in healthy, younger populations may become a valuable national strategy.

Correspondence: Renata J. M. Engler, MD, Allergy-Immunology Department/Vaccine Healthcare Centers Network, Walter Reed Army Medical Center, 6900 Georgia Ave NW, Bldg 41, Ste 21, Washington, DC 20307 (renata.engler@gmail.com) (renata.engler@amedd.army.mil).

Accepted for Publication: April 14, 2008.

Author Contributions: Dr Engler had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Engler, Nelson, Klote, VanRaden, Cox, Klimov, Keitel, and Nichol. Acquisition of data: Engler, Nelson, Klote, Cox, Klimov, and Carr. Analysis and interpretation of data: Engler, Nelson, VanRaden, Huang, Cox, Klimov, Keitel, Nichol, Carr, and Treanor. Drafting of the manuscript: Engler, Nelson, VanRaden, Keitel, and Carr. Critical revision of the manuscript for important intellectual content: Engler, Nelson, Klote, VanRaden, Huang, Cox, Klimov, Keitel, Nichol, Carr, and Treanor. Statistical analysis: Engler, VanRaden, and Huang. Obtained funding: Engler, Cox, Klimov, and Carr. Administrative, technical, and material support: Engler, Nelson, Cox, Klimov, Keitel, and Carr. Study supervision: Engler, Nelson, Klote, Klimov, and Carr.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the Office of the Army Surgeon General in collaboration with Walter Reed Army Medical Center (WRAMC) and Healthcare System; the North Atlantic Regional Medical Command; the US Army Medical Research and Materiel Command; the National Institute of Allergy and Infectious Diseases, National Institutes of Health; and the Influenza Branch of the Centers for Disease Control and Prevention.

Disclaimer: The views expressed in this article are those of the authors and do not reflect the official policy of the Department of the Army, Department of Defense, US government, National Institutes of Health, Centers for Disease Control and Prevention, or other federal agencies.

Walter Reed Health Care System Influenza Vaccine Consortium: Bryan L. Martin, DO (Colonel, US Army, retired); Denece Shelton, RN, BSN; Ha Tran, BS (Captain, US Army) and the WRAMC Pathology Department; Bruce McClenathan, MD (Major, US Army); Margaret Yacovone, MD (Lieutenant Colonel, US Army); Ronald DeGuzman, MD (Lieutenant Colonel, US Army); Karla Davis, MD (Major, US Army); Cecilia Mikita, MD (Major, US Army); Ann Desoto, BA; Limone C. Collins, MD; John J. Moore, MD (Colonel, US Army, retired); and the dedicated staff of the Walter Reed Regional Vaccine Healthcare Center.

Additional Contributions: We acknowledge and give special thanks to the following individuals and organizations for the remarkable interagency rapid response and teamwork support of this research effort, developed from an idea at the National Vaccine Advisory Committee on October 10, 2004, to full-protocol approval (with the Food and Drug Administration–approved investigational new drug) and implementation by November 8, 2004: Kevin Kiley, MD (Lieutenant General, US Army, retired), the Army Office of the Surgeon General, without whom this study could never have been done; Harold Timboe, MD (Major General, US Army, retired), and Kenneth Farmer, MD (Major General, US Army, retired), as Commanders of the North Atlantic Regional Medical Command; the dedicated staff of the Walter Reed Army Medical Center Department of Clinical Investigation, particularly Maria Sjogren, MD (Colonel, US Army, retired), Susan D. Fracisco, MD (Colonel, US Army), and the investigational review board (IRB) and human use committee (HUC); Walter Reed Health Care System Influenza Vaccine Consortium; George Curlin, MD, Division of Microbiology and Infectious Diseases, National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health; Qian Dong, NIAID (statistics); Lester Martinez-Lopez, MD (Major General, US Army, retired); Jerome F. Pierson, RPh, PhD (Colonel, US Army, retired), Shirley Roach, and Ann Altman (Lieutenant Colonel, US Army, retired), US Army Medical Materiel Development Activity; Leon Moore and Tony Chen from Uniformed Services University of the Health Sciences Center for Informatics in Medicine for their support of the online survey questionnaire; John Grabenstein (Colonel, US Army, retired), Allison R. Christ (Captain, US Army), and the staff of Military Vaccine Agency; Dale Block (Colonel, US Army), Pentagon DiLorenzo Clinic, and his staff; Walter Reed Army Institute of Research IRB; Walter Reed Department of Pathology; the dedicated nursing staff of the Allergy-Immunology Department, WRAMC, particularly Sadie H. Massey, Norma S. Veltri; Fran Lessans and the staff at Passport Health; McKesson Biosciences, Rockville, Maryland, for rapid response labeling; and Sandii Mon, Gerard Reardon, PhD, and Christina Spooner, BS, MS, from the Vaccine Healthcare Centers Network.